Introduction
Vertical Migration
Vertical
migration by zooplankton is one of the earliest and most commonly observed
behavioral phenomena in both marine and freshwaters (Cuvier, 1817; Leydig,
1862; Russell, 1927). Alternately
referred to as diel or diurnal vertical migration (DVM), this usually
involves a migration upward from deeper waters at dusk, resulting in a
population maximum in relatively shallow water at night, followed by
sinking to deeper water at dawn and a mid-day population maximum at depth.
Deviations from this ‘nocturnal’ DVM include ‘reverse’ DVM,
with a single surface maximum during the day, and a ‘twilight’ DVM in
which surface maxima form both at dawn and dusk (Hutchinson, 1967; for
examples, see: Dumont, 1972; Cunningham, 1972; Ohman et al., 1983).
The
proximal cause of migration has most often been found to be relative
changes in light intensity (Siebeck, 1980; McNaught and Hassler, 1964;
Ringelberg, 1964), while the ultimate reason for this behavior is somewhat
more obscure. The most common
pattern of nocturnal migration involves removal of the animals from a
warm, food-rich environment at night to a cold, food-poor environment
during the day, at a substantial energetic cost.
Most attempts to explain this apparent paradox fall into one of two
categories (Lampert, 1989):
-
Vertical
migration provides a metabolic or demographic advantage, or
-
Avoidance
of surface waters during the day reduces losses from predators.
The
first hypothesis, originally proposed by McLaren (1963; 1974) assumes
either that a resting period in low temperatures confers a long-term
energetic advantage, or that low temperatures allow individuals to reach a
larger adult body size, which in turn results in greater fecundity.
The second hypothesis, which is probably more widely accepted,
holds that migration to a dark refugium in deep waters during daylight
hours reduces losses due to visual predators (Zaret and Suffern, 1976).
Within
a lake, the pattern and amplitude of DVM can vary from species to species
(Wells, 1960; Geller, 1986; Angeli et al., 1995) and within a species with
size, age and sex (Stich, 1989). In
general, more visually conspicuous individuals (e.g., larger animals,
gravid females) exhibit greater amplitudes of migration (Wright et al.,
1980; Haney and Hall, 1975). The
strength and extent of DVM can also vary seasonally for a given species
within the same habitat (Stich and Lampert, 1981; Ringelberg et al.,
1991). Often in deep lakes
migratory behavior begins soon after the clear water phase in June
(Geller, 1986; Stich, 1989). Migratory
behavior can vary for a given species from year to year, and from lake to
lake, in response to changes in predator abundance.
Williamson and Magnien (1982), Stirling et al. (1990), Lehman and
Caceres (1993) and Frost and Bollens (1992), have all found interannual
variations in the amplitude of zooplankton migration to be related to
changes in the size of predator populations, while Gliwicz (1986) found a
relationship between amplitude of migration of copepods and the age of the
fish populations across a number of lakes that had been stocked with char
for different periods of time.
Brief
History of the GLNPO Zooplankton Sampling Protocol
The Great Lakes National Program Office (GLNPO) began annual monitoring of
the Great Lakes in 1983 for Lakes Michigan, Huron, and Erie.
In 1986 sampling was extended to include Lake Ontario, and in 1992
sampling of Lake Superior was added.
In 1983 and 1984, two zooplankton tows were taken at each site with
a 62 mm
mesh net: one from 2 m above the bottom to the surface, and a second from
20 m to the surface, although little analysis was apparently conducted (Makarewicz,
1987; Makarewicz, 1988). In
1985, the deep tow seems to have been discontinued (Makarewicz and
Bertram, 1991).
Prior
to the summer, 1997 cruise, the zooplankton sampling protocol was changed
to include 100 m tows, in addition to the 20 m tows.
Unlike previous deep tows, the 100 m tows were taken using a net
with a larger mesh size (153 µm) to prevent clogging and to reduce the
pressure wave created by the net during sampling.
Also, time of day the tows are taken was recorded from 1996 on.
The
depth at which zooplankton tows are taken can have a number of possible
effects on the estimation of zooplankton abundance and community
composition. In species that exhibit nocturnal DVM with a maximum depth
greater than 20 m, shallow tows taken during the day can result in an
underestimation of abundances. In
addition, if larger animals exhibit greater amplitudes of migration than
smaller animals, then 20 m tows will have a disproportionately large
percentage of smaller animals during the day, and thus individual and
total biovolumes of migrating animals will be underestimated in day
samples. These two factors
can result in the appearance of spurious horizontal spatial patterns in
zooplankton community data where in fact none exist, particularly if sites
near to each other are sampled at similar times of day, as is often the
case. Additionally,
populations of deeper-living zooplankton that rarely migrate above 20 m
would be consistently underestimated in 20 m tows, whether taken during
the day or at night. On the
other hand, the abundances of animals whose maximum depth of occurrence is
substantially above 100 m would be underestimated, on a volumetric basis,
by deeper tows, due to ‘dilution’ of the population by the sampling of
deep water which is essentially free of any individuals.
Here
we examine the comparability of data collected with the two different
tows. In particular, we
sought to address the following questions:
-
Are
there differences in the net efficiencies of the two mesh sizes (64
µm and 153 µm) to capture specific zooplankters?
-
Do
shallow tows result in differences in relative community composition
during the day? If so,
what species’ abundances are underestimated during the day by
shallow tows?
-
Are
there lake to lake or within lake differences in the suitability of
shallow tows to estimate abundances of specific species?
-
Are
there species which are underestimated at all times by shallow tows?
-
Do
shallow tows result in low bias in length measurements (and hence
biovolume estimates) during the day?
-
Are
volumetric estimates of abundance and biovolume underestimated by
deep tows?
Methods
Sample Collection
and Processing
Samples
used in this study were collected during the summer, 1998 cruise, during
which a total of 72 stations were sampled for zooplankton
(Figure
1).
Two sampling tows were performed at each site, using a 0.5 m
diameter conical net (D:L = 1:3). The
first tow was taken from 20 meters below the water surface using a 63 µm
mesh net, and the second tow from 2 meters above the bottom of the lake or
100 m, whichever was less, using a 153 µm mesh net. If
the station depth was less than 20 m, both tows were taken from one meter
above the bottom. Triplicate
tows of each depth were taken at the Master Stations.
After
collection, samples were immediately narcotized with soda water, and were
preserved with sucrose formalin solution (Haney and Hall, 1973)
approximately twenty minutes later. Samples
were split in the lab using a Folsom plankton splitter, and four
stratified aliquots examined per sample.
Length measurements were made on the first twenty individuals of
each species encountered per sample.
Identifications followed Balcer et al. (1984) for adult calanoids,
malacostracans, the cladocerans Leptodora
kindtii, Polyphemus pediculus, Holopedium
gibberum, and Diaphanosoma
birgei; Hudson et al. (1998) for adult cyclopoids and harpacticoids.
Brooks (1957) and Evans (1985) were used for all Daphnidae, and the
remaining cladocerans (Chydoridae, Bosminidae, and Macrothricidae) were
classified according to Edmundson (1959).
Members of Cercopagidae (i.e. Bythotrephes
cedarstroemi, Cercopagis pengoi)
were identified according to Rivier (1998).
Analytical Approach
To
determine the comparability of data generated from nets with the two
different mesh sizes (64 µm and 153 µm), all other factors being equal,
sites less than 20 m (i.e. where tows with both nets were taken from the
same depth) were examined. There
were 12 such sites in the western and central basins of Lake Erie during
the summer, 1998 cruise. For
each species, estimates of abundances (# m-3) generated from
the two different tows were compared to determine if there were any
changes in collection efficiency that could be attributed to differences
in net mesh size. Comparisons
were made using the Wilcoxon signed rank test, a nonparametric
paired-sample test which is used to examine data in which measures are
repeated only once (e.g. replicate tows with different mesh sizes).
The use of a paired-sample test in this case eliminated the
influence of inter-site variability on the analysis.
Count data tends to follow a Poisson, rather than a normal,
distribution, and thus violates the normality assumption of parametric
tests.
Differences in relative community composition estimates as a result
of different tow depths were assessed using Whittaker’s (1952) percent
similarity (PSC) index:
where
a and b are, for a given species, percentages of the total samples A
and B which that species represents.
The absolute value of their difference is summed over all species.
This number ranges from 0 to 1, with 0 indicating two samples with
no species in common, and 1 indicating two samples with all species
present in both samples in the same relative proportions.
A relative index, rather than an absolute index (e.g. Pinkham and
Pearson’s (1976) Biosim index) was used to disregard effects of dilution
in cases where the majority of individuals were present substantially
above 100 m.
If a substantial portion of the zooplankton community underwent
diel migration to depths greater than 20 m, then one would expect
similarities between shallow and deep tows to be greater for sites visited
at night than for sites visited during the day.
This was tested for each lake by first classifying samples
according to whether they were collected during the day or at night, where
day was defined as being from one hour after sunrise to one hour before
sunset. Percent similarities
of zooplankton communities estimated with deep and shallow tows were then
calculated for each site (in the case of Lake Erie, for the eastern basin
only), and differences in similarities between day samples and night
samples assessed with a t-test
where PSC values were both normally distributed and exhibited
homoscedasticity, and with a Wilcoxon rank sum test where the assumptions
of normality and homoscedasticity were not met.
To determine if individual species were undergoing migration,
two-way nested ANOVAs were conducted on species abundances within each of
the upper lakes. Here, depth
of tow (Depth) was one factor, and time of day (Time), e.g. day or night,
the other factor, and the response variable was species abundance m-3
for each species within each lake. To isolate the variation due to site differences, Site was
included as a factor in the analysis.
Since each site was only sampled during one time period (i.e.
either day or night), but two tows were taken at each site, factor effects
of Site were crossed with depth of tow, but nested within time of day (Figure
2). If species were
undergoing DVM of an amplitude that took them substantially out of the
upper 20 m during the day, then the magnitude of difference in numbers of
individuals estimated from deep and shallow tows would be dependent on the
time of day the tows were taken. This
would show up as a significant interaction effect between the factors
Depth and Time. Since site
was nested within time, and its variance was consequently used as the
denominator in tests of significance for factors including time, high site
to site variance for a given species within a lake would result in reduced
power of the test to detect both Time effects and migration (i.e. Time x
Depth interaction) effects. A significant effect of Depth, without a significant
interaction effect, could be due to either a preference of the species for
depths substantially above the depth of the deep tow, and hence dilution
of the organism in the deep tow, or a preference for depths greater than
20 m, and therefore an underestimation of abundance by the 20 m tow.
Separate ANOVAs were conducted for each species within each lake,
and the data were assessed for conformance to the assumptions of normality
and homoscedasticity. Where these assumptions were not met, one of the following
transformations was used, depending on which stabilized variance and
resulted in a normal distribution:
Sites
in both the western basin and the central basin of Lake Erie are
relatively shallow, with mean depths of 8.8 m and 25.5 m, respectively.
These sites were therefore not included in these analyses.
The eastern basin of Lake Erie is deeper, with an average site
depth of 47.5 m.
However, there were only four sites visited in this basin, three
during the day and one at night. Similarly, in Lake Ontario species distributions were uneven
enough across the lake to limit the number of sites that could be used for
any one ANOVA analysis. As a
result, a different approach was used for these two lakes. Here, differences in the abundances of individual species
between deep and shallow tows for sites visited during the day were tested
for using a paired test. Assumptions
of homoscedasticity and normal distribution were met in all cases, so a
one-way paired t-test was used.
The null hypothesis tested was:
where:
= the average of the differences in abundances between shallow and deep
tows at each station. If
abundances were greater in deep tows than in shallow tows at stations
sampled during the day, this would suggest the possibility that animals
were migrating below 20 m during the day.
To compare differences in length between animals captured by the
shallow tows and animals captured by the deep tows, average lengths for
deep daytime and shallow daytime tows were compared within each lake using
either a t-test, or the
Mann-Whitney rank sum test in the event of non-normality or
heteroscedasticity. With this
approach, variability due to site to site differences was confounded with
variability due to differences in length as a result of depth.
However, the considerable non-normality of most the data, and the
lack of an appropriate nonparametric method to incorporate site to site
differences, particularly given the large differences in sample size
within each cell, precluded a better approach.
It should be borne in mind, therefore, that the power of this test
is probably quite low.
Results
Differences
in Net Efficiency Between 64 and 153 mm Mesh Nets
A
total of 24 crustacean taxa were found at the 12 sites with depths less
than 20 m. Of these, only one taxa, Mesocyclops
copepodites, showed a significant (a
= 0.05) difference between the abundance estimates made with the two
different mesh sizes (Table 1). At
an a
of 0.05, it is expected that one type I error will be committed, on
average, for every 20 analyses. Therefore,
there appears to be no real difference in the estimates of
macrozooplankton abundances collected from a fixed depth using a 64 µm
compared to a 153 µm mesh net.
Differences
in
Relative
Community
Composition
Between
Deep
and Shallow
Tows
Zooplankton
community composition by major taxa, as estimated using deep and shallow
tows, are shown for all five lakes in Figures
3,
4,
5,
6 and
7.
In Lake Superior, nine of 19 stations were sampled during the day
(Figure 3). All sites were
dominated by immature copepods. While
volumetric abundances differed between shallow and deep tows taken during
the day, differences in relative community composition, at least on the
basis of broad taxonomic groups, were not apparent.
In both Lakes Michigan and Huron (Figures 4 and 5), about half of the
tows were taken during the day. When
shallow tows taken during the day were compared with those taken at night,
clear differences in both species composition and abundance where
apparent, with day tows showing consistently lower volumetric abundance
estimates and a greater proportion of immatures compared to night tows.
Deep tows in both lakes showed a remarkable degree of spatial
homogeneity, in terms of both species composition and total abundance. In Lake Erie, sites in the western and central basins were by
and large 20 m or less in depth, so not surprisingly no substantial
differences were noted in community composition between deep and shallow
tows taken during the day (Figure 6). On
the other hand, in the eastern basin, the genus Bosmina,
which was dominant in the deep tows, was almost completely missing from
all shallow tows. Lake
Ontario, like Lake Erie, exhibited a notable degree of spatial
heterogeneity in community composition, with the western three sites
dominated by Bosmina, and the eastern four sites dominated by Daphnia
(Figure 7). A central site
was intermediate in species composition between these two communities.
Half of the sites were visited during the day, and no systematic
differences in community composition were apparent, although abundances
tended to be higher in the shallow tows.
To
determine if differences in community composition from shallow and deep
tows were statistically significant for each lake, PSC similarity values
between shallow and deep tows for each site collected during the day were
compared to similarity values between shallow and deep tows collected at
night (Figure 8). In Lake Erie, only sites in the eastern basin were compared,
since sites in the western and central basins were not substantially
deeper than 20 m. Statistically
significant differences (a = 0.05) were found for
Lakes Huron and Michigan, indicating that shallow and deep tows taken at
night were more similar to each other than shallow and deep tows taken
during the day. This implies
that a substantial portion of the zooplankton community was undertaking
nocturnal DVM to a depth greater than 20 m.
Although a very large difference in the similarity of deep and
shallow tows between day and night was apparent in the eastern basin of
Lake Erie, the small sample size precluded a statistically significant
difference from being detected. No
differences in similarity between day and night samples were detected in
Lake Ontario, where similarity values were uniformly high. In Lake Superior, similarities between shallow and deep tows
were comparable for day and night samples; however, overall similarities
between shallow and deep tows were lower than for other the lakes.
This implies that there were differences in the depth distributions
of species both day and night.
Species-Specific
Differences
in
Depth
Distribution
and
Migration
Behavior
To
test for depth preferences and indications of migratory behavior on a
species by species basis, ANOVA analyses were conducted on individual
species from Lakes Huron, Michigan and Superior.
The full results of all ANOVA analyses are presented in
Table
2.
Cladocera:
There were five species of cladocera present in sufficient numbers
in at least one of the upper lakes to permit analysis: Bosmina
longirostris, Eubosmina coregoni, Holopedium
gibberum, Daphnia galeata
mendotae and Bythotrephes cedarstroemi.
Abundances of B. longirostris were high
enough in both Lakes Huron and Michigan to permit analysis.
In both lakes there was a significant interaction effect between
Time and Depth, indicating the likelihood of DVM.
This can be seen in box plots of abundances of this organism, which
show that shallow tows taken at night result in significantly higher
abundance estimates than shallow tows taken during the day for both lakes
(Figure 9). Abundances of E.
coregoni in Lake Huron were substantially greater in both deep day and
shallow night samples, compared to shallow day samples (Figure
10),
suggesting nocturnal DVM, although there were no significant factor
effects. Failure to find
significant differences could have been a result the low power of the
performed test, however (b
= 0.076 for Time x Depth). H.
gibberum, present in Lake Superior, showed significant Depth effects,
with abundances estimated from shallow tows consistently higher than those
from deep tows (Figure 11). This
suggests that H. gibberum prefers upper waters, and therefore that deeper tows
consistently underestimate its abundance.
There was no evidence of migration in this species.
D. galeata
mendotae was present in substantial numbers in all three upper
lakes, although its behavior seemed to differ from lake to lake.
In both Lakes Michigan and Huron, significant Depth x Time
interaction effects were found, suggesting that D.
galeata mendotae undergoes DVM in these lakes (Figure
12).
This was particularly pronounced in Lake Michigan, where mean
abundances in shallow night tows were nearly two orders of magnitude
higher than in shallow day tows. In
both lakes, deep tows were greater than day shallow tows, and less than
night shallow tows, indicating that though animals were migrating above
and below 20 m, most of the population was well above the depth of the
deep tows, and so abundances were underestimated by the deep tows.
In Lake Superior, in contrast, no significant interaction effects
were found, but Depth had a significant effect on abundance estimates.
Mean abundance estimates from shallow tows were significantly
higher than those from deep tows, indicating that animals were staying
above 20 m both day and night, and consequently deep tows underestimated
abundance by diluting the samples. B. cedarstroemi was common
in Lakes Huron and Superior; no evidence of migration below 20 m was found
in either lake (Figure 13).
Paired t-test analyses were conducted on the species abundances estimated
from deep and shallow tows at the stations sampled during the day in Lakes
Erie and Ontario. Full
results from these analyses are presented in
Table 3.
In Lake Erie, a statistically significant difference between
abundance estimates from deep and shallow tows was found only for Bosmina longirostris. Interestingly,
a difference was not found for this organism in Lake Ontario.
When the relative abundances of this species in the two lakes are
examined (Figure 30), dramatic differences in its vertical distribution are
apparent. The relative
abundances of B. longirostris
estimated from deep and shallow tows at sites in Lake Ontario sampled
during the day are essentially identical, while in Lake Erie this organism
is almost completely absent from the upper 20 m during the day at sites in
which it appears to be the dominant organism, judging from abundances
estimated from the deeper tows.
Copepoda:
Of the copepods, there were two species of cyclopoids and seven
species of calanoids abundant enough in at least one lake to test.
In addition, immatures were tested.
The cyclopoid Tropocyclops
prasinus mexicanus was found in Lake Michigan, and both adult and
immatures were significantly more prevalent in the upper 20 m both day and
night, with no evidence of migration below 20 m (Figure
14).
Both mature Diacyclops thomasi and its
copepodites showed evidence of extremely strong DVM in Lake Michigan, with
adult abundances in shallow night tows about two orders of magnitude
higher than shallow day tows (Figure 15).
The extremely low abundances found in the shallow day tows strongly
suggest that most of the adult population was migrating below 20 m during
the day. In Lake Superior on
the other hand, there was no indication of migration below 20 m during the
day for this species (Figure 16). In
fact, both adults and copepodites were more abundant in shallow day tows
than in shallow night tows, although this difference was not statistically
significant. There was a
statistically significant effect of Depth, pointing to dilution of the
deep tow samples. D.
thomasi populations in Lake Huron showed no statistically significant
effects of Depth or Time (Figure 17).
Calanoid copepods were the most diverse group of the
macrozooplankton, with a total of seven species found.
Four species of the family Diaptomidae were found in the lakes: Leptodiaptomus
ashlandi, Leptodiaptomus minutus,
Skistodiaptomus oregonensis
and Leptodiaptomus sicilis. Abundances of L. ashlandi
from shallow tows in both Lakes Huron and Michigan were several times
higher at night than during the day, strongly suggesting DVM (Figure
18).
However, these differences were not statistically significant.
A similar, but stronger, pattern was found in both lakes for L.
minutus, indicating strong DVM
below 20 m during the day (Figure 19).
Deep tows were similar both day and night, and were lower than
shallow night tows and higher than shallow day tows.
This indicates that most of the population was probably well above
100 m at all times. Substantial
populations of S. oregonensis
were found only in Lake Michigan, and while shallow night tows were much
higher than shallow day tows, there was no statistically significant
difference (Figure 20). Since
deep tows were always less than shallow tows, indicating dilution of these
population in deep tows, it appears that both shallow day and all deep
tows underestimate these populations.
L. sicilis was present at
a substantial number of sites in all three lakes, and its populations were
always significantly greater in deep tows than in shallow tows (Figure
21).
There was some suggestion of migration into the upper 20 m at night
in Lakes Huron and Michigan, although this was not statistically
significant. Percent
abundance data of L. sicilis
clearly indicate that shallow tows greatly underestimate abundances of
this species in Lake Michigan and Superior (Figure 22).
Diaptomid copepodites showed evidence of migration in Lake
Michigan, and of increased numbers with depth in Lake Superior (Figure
23).
No patterns were noted in Lake Huron.
Since this group probably represented a mix of species with
different behaviors, however, conclusions regarding differences between
lakes are of limited value.
Limnocalanus macrurus
was found in both Lakes Michigan and Superior, and in both lakes relative
abundances were significantly higher in deep tows, both day and night,
than shallow tows taken at any time (Figure 24).
In addition, animals were nearly absent from the upper 20 m during
the day, but not at night, suggesting migration into the upper 20 m at
night (Figure 25), although a significant interaction effect was noted only
in Lake Superior. The lack of an interaction effect in Lake Michigan was
probably due to the high variability in that lake.
In both lakes, abundances estimates were higher from deep night
tows than deep day tows, suggesting that perhaps a portion of the
population was residing in waters deeper than 100 m during the day and
migrating up at night. Senecella calanoides, also
present in both Lakes Superior and Michigan, showed an even more extreme
distribution, being completely absent from surface waters both day and
night (Figure 26). Somewhat
more individuals were found in deep night tows than in deep day tows,
suggesting again that perhaps some of the population was migrating below
100 m during the day, but numbers of individuals were too low to permit
confident conclusions. As
with L. macrurus, estimates of
the relative abundance of S. calanoides made with deep and shallow tows were very different (Figure
27).
Epischura
lacustris was present at a substantial number of sites in all lakes,
and showed some evidence of migration in Lake Michigan, but not in the
other lakes (Figure 28). Epischura
copepodites, present both in Lake Michigan and Lake Huron, were
significantly more abundant in the upper 20 m in both lakes (Figure
29).
In
Lake Ontario, diaptomid copepodites showed a slightly significant
difference in abundances estimated from the two tows (Table 3).
None of the other copepods tested in the two lower lakes exhibited
any differences in vertical distribution, indicating that DVM below 20 m
was not occurring for these species.
Differences in
Animal
Lengths
Between
Deep and
Sallow
Tows
Full results of length comparisons between shallow and deep day
tows are presented in Table 4. Of
the twenty species tested, eleven showed significant differences in length
between individuals captured in the deep tows compared to the shallow tows
in at least one lake. Only in the case of D.
thomasi were larger individuals found in the shallow tows.
Differences in length, though significant, were in most cases
relatively minor, often amounting to less than a one percent change in
lengths between the two tows. However,
for some organisms, in particular the cladocerans B.
longirostris and D. galeata mendotae, the differences were substantial.
In the case of Lake Huron, differences in length of Bosmina
longirostris and D. galeata
mendotae between shallow and deep tows would have lead to an
underestimation in biovolume of approximately 50%, even if abundances
calculated from the two tows were identical.
Discussions
Migration
A
number of different patterns of depth distribution were found for the
zooplankton in the Great Lakes. In
general, cladocerans exhibited nocturnal DVM in Lakes Michigan and Huron,
with patterns being stronger in the former lake.
B. longirostris, the
dominant cladoceran in eastern Lake Erie, also appeared to be undergoing
migration in that lake, being virtually absent from 20 m tows during the
day. In contrast, where
sufficient organisms were present to test, migratory behavior was not
apparent in cladoceran populations in Lake Superior or Lake Ontario.
In Lake Superior in particular, cladocerans exhibited a strong
tendency to remain in the upper 20 m both day and night.
Finding
comparable data in the literature is difficult since most lakes are much
shallower than the Great Lakes, and therefore migration patterns tend to
be much more restricted with respect to depth.
Surprisingly, little work on migratory behavior has been done on
the Great Lakes, although some studies have been conducted which have
addressed vertical distribution. Wells
(1960), in an extensive study of the zooplankton of Lake Michigan, found
that D. galeata mendotae
undertook diel migrations, but that populations did not descend below 20
m. In contrast to the present
study, his data indicated that migration took place almost entirely
between the surface and 10 m. McNaught
and Hasler (1966), on the other hand, found that populations of D.
retrocurva in Lake Michigan had an amplitude of migration of over 20
m, with population mean density at a depth of 34 m during the day.
Conway et al. (1973) in a study of zooplankton distribution in Lake
Superior found that abundances of cladocerans declined notably between 20
and 30 m, which is consistent with the present findings.
Wilson and Roff (1973) observed variable migration in Bosmina
in Lake Ontario. During
September there was notable migration below 20 m during the day, although
the majority of the population remained above that depth.
In contrast, D. retrocurva
exhibited stronger migration below 20 m in their study.
Migration ranges for the two species in September were about 20 m
and 14 m, respectively.
In
Lake Geneva, Angeli et al. (1995) found that migration of small
individuals of D. hyalina was
limited to the upper 15 m, while larger individuals undertook migrations
with an amplitude of greater than 30 m. Makino et al. (1996), working on a caldera lake in Japan,
observed population maxima of D.
longispina between 25 – 50 m during the day in spring, but mostly
above 30 m in October. Similar
depth maxima were found for Bosmina
(Eubosmina) coregoni.
In contrast to the smaller cladocerans, the large invertebrate
predators Leptodora kindtii, Bythotrephes
cedarstroemi and Cercopagis
pengoi did not exhibit substantial migratory behavior in any of the
lakes. This seems
paradoxical, since the size of these zooplankton would appear to make them
more vulnerable to sight-feeding planktivores.
However, Leptodora kindtii
is extremely transparent, and it is possible that even in epilimnetic
waters during the day it is not visible to fish.
Both Bythotrephes and Cercopagis have very long barbed spines which can get caught in the
gill rakers of planktivorous fish (Barnhisel and Harvey 1995), thus
reducing their vulnerability to predation.
Stich (1989), working on Lake Constance, also found that migratory
behavior of the species Leptodora
kindtii and Bythotrephes
longimanus was either slight or absent.
Our study also found evidence of migration in many copepod species,
although in some cases site to site variability prevented this from being
statistically significant. Of
the cyclopoids, only D. thomasi showed evidence of migration, and then only in Lake
Michigan. On the other hand,
most of the calanoids, and in particular the diaptomids, exhibited depth
distributions in Lakes Michigan and Huron suggestive of migratory
behavior, although in some cases these were not statistically significant. Evidence of migration of these organisms was not apparent in
the other lakes. Wells (1960)
similarly found nocturnal DVM, extending to below 20 m, for the diaptomids
L. minutus, L. ashlandi and S. oregonensis.
No evidence of migration was found for these organisms in the lower
lakes, which agrees with Wilson and Roff’s (1973) observations on Lake
Ontario.
There
are a number of possible explanations for the differences observed in
migration patterns from lake to lake.
The lack of observable migration in Lakes Superior and Ontario
could be due to a smaller amplitude of migration in those lakes.
If movements were largely confined to the top 20 m, migratory
behavior would not have been apparent from our analyses.
These differences could also be due to differences in predation
pressure by juvenile or small fish. There
is a great deal of literature documenting the influence of the presence of
predators on the extent of DVM in freshwater zooplankton.
In field studies of Polish lakes, Gliwicz (1986) found that Cyclops abssorum migrates only where Arctic char are present, and
this behavior can be induced through planktivorous fish stocking.
The initiation of vertical migration in Daphnia
hyalina in Lake Maarsseveen has been observed to coincide with the
appearance of active juvenile perch, and to cease when these predators
disappeared from the open water (Ringelberg et al., 1991).
Similarly, migration of Diaptomus kenai in Gwendoline Lake, British Columbia ceased when the
predator Chaoborous was removed
from the lake by an invasion of small trout, but when Chaoborous were added in an in
situ enclosure, DVM resumed within 4 hours (Neill 1990).
Stirling et al. (1990) found that DVM of Daphnia
galeata mendotae in Lake St. George increased in amplitude by a factor
of 2 during years in which planktivorous fish recruitment was high, while Daphnia
pulex responded to Chaoborous
additions in Ranger Lake, Ontario by initiating migratory behavior
(Nesbitt and Riessen, 1996).
Laboratory
work has provided evidence that the induction of DVM in zooplankton is a
response to the presence of chemical exudates produced by predators.
Dodson (1988) first demonstrated that several different species of Daphnia
could be induced to ascend or descend when added to water that was
preconditioned by the presence of a vertebrate or invertebrate predator,
and that this response was both predator- and prey-specific.
This initial observation has been corroborated by a number of
subsequent laboratory studies (e.g., Ringelberg, 1991; Loose, 1993; Watt
and Young, 1994; Van Gool and Ringelberg, 1998), and recent attempts have
been made to chemically characterize the fish exudates responsible for
this behavior (Von Elert and Loose, 1996).
The response of zooplankton exposed to such predator exudates is
typically rapid (Neill, 1990) but short-lived (Dodson, 1988; Loose, 1993).
In general, contact with vertebrate predators, which are mainly
visual feeders, induces ‘normal’ DVM, while invertebrate predators,
which are mainly tactile feeders and often undergo normal DVM themselves,
induce reductions in mean population depths or reverse migrations (Leibold,
1990; Nesbitt and Riessen, 1996; Dodson, 1988; Brancelj and Blejec, 1994).
In this way contact time with the two predator groups is reduced.
The
intensity of response to predator cues appears to have a genetic component
(Young and Watt, 1994), and it has been supposed that this serves to set a
limit on the maximum extent of behavioral response to predator cues.
So populations from environments historically subject to greater
predation pressure might be genetically predisposed to exhibit greater
behavioral responses to predator fish cues than those populations which
have been subject to less intense pressure (cf. Young and Watt, 1996).
These genetic differences can even be exhibited within a given
habitat between genetically distinct populations of a single species. Field studies have found distinct migratory behaviors between
co-occurring but genetically distinct populations of Daphnia longispina (King
and Miracle, 1995), Daphnia galeata
x hyalina (Spaak and Ringelberg, 1997; Van Gool and Ringelberg, 1998)
and Daphnia pulex (Weider, 1984).
It
is possible that the differences in migratory behavior observed in the
present study could be the result of differences in predation pressure
between the lakes. This would
suggest that predation pressure by zooplanktivorous fish is greater, for
example, in Lake Erie, where migration by Bosmina longirostris was observed, than in Lake Ontario, where such
behavior was not apparent. Similarly,
zooplanktivory by vertebrates might be expected to be more intense in
Lakes Michigan and Huron than in Lake Superior.
Dorazio et al. (1987) have reported differences in the migratory
behavior of Daphnia species in
Lake Michigan that they attributed to interannual differences in
planktivore populations. In
1985, during which the planktivore Coregonus
hoyi was present in large
numbers, all zooplankton grazers displayed pronounced vertical migrations,
while in 1983, a year of low planktivore numbers, the dominant herbivore, D. pulicaria, did not exhibit vertical migration.
The lack of apparent migration in Lakes Superior and Ontario, on
the other hand, might be the result of predation pressure by
invertebrates. The main
invertebrate predator in the lakes, Bythotrephes
cedarstroemi, was present at most sites in Superior, albeit at very
low abundances. However,
populations of this predator were much higher in eastern Erie than in
western Ontario, which is the opposite of what would be expected on the
basis of the migration patterns of B.
longirostris in these two areas.
Alternatively, nocturnal diurnal migration by planktivores can
result in static populations of zooplankton in shallow waters.
Levy (1990) has shown that Bosmina
populations remained in shallow waters in British Columbia lakes where
juvenile sockeye salmon migrated vertically, but undertook vertical
migrations where sockeye vertical migrations were reversed and
sticklebacks were present. More
concrete conclusions about the possible causes for the differences in
vertical distributional patterns observed in our study will have to await
an analysis of fish census data in the lakes.
Depth
Distribution
A number of calanoid species were found to exhibit depth
preferences that resulted in a majority of the population residing below
20 m at all times. Limnocalanus
macrurus, present at a
substantial number of sites in both Lakes Michigan and Superior, was
significantly more abundant in the 100 m tows, both day and night, while Senecella calanoides, also
present in both lakes, was virtually absent from the 20 m tows at all
times. Leptodiaptomus sicilis,
which was present at most sites in all three upper lakes, was also always
more abundant in the deeper tows, although the difference in abundance
between deep and shallow tows was not as pronounced as for the other two
species.
Other studies have documented the deep water preferences of these
species. In Wells’ (1960)
study of Lake Michigan, L. macrurus was almost completely absent from the top 20 m during a
period of strong thermocline delineation, but when thermal structure was
weaker animals were found in surface waters at night, suggesting that a
sharp temperature gradient could restrict its movements. McNaught and Hasler (1966) also found evidence that a sharp
thermal gradient presented a hindrance to L.
macrurus movement in L. Michigan.
Wilson and Roff (1973) reported a mean depth of 50-62 m for the L.
macrurus population in Lake Ontario.
While usually below 20 m, the animals did reach the surface waters
at night through the summer and fall, although apparently a distinct
epilimnion did not form during their study.
Conway et al. (1973), working in Lake Superior, found that L.
macrurus was present at all depths from June through early August, but
was restricted to depths below the thermocline when it was present. L. macrurus
appeared to be restricted to a temperature below 12°
C. Wells (1960) found that
abundances of S. calanoides in
Lake Michigan were usually highest below 20 m.
This species was rare in his collections, but since his tows only
extended to 40 m, it is possible that the majority of its populations were
missed. Conway et al. (1973)
rarely found this animal above 40 m in Lake Superior.
It is clear, therefore, that shallower tows run the risk of
substantially underestimating the abundances of these species, or missing
them altogether.
Lengths of
Migrating
Individuals
In Lakes Michigan and Huron where migration was occurring, the
larger cladocerans appeared to exhibit greater amplitudes of migration
than the smaller organisms. This phenomenon was first suggested by Wilson and Roff (1973)
when they suggested a relationship between zooplankton body weight and
range of migration. Haney and
Hall (1975) found that for Daphnia
pulex and Daphnia galeata
mendotae, filtering activity at the surface at night was up to 25x
higher due to the presence of larger-bodied animals, which were absent
from surface waters during the day. Wright
et al. (1980) used a model to predict that larger individuals and gravid
females of Daphnia parvula would exhibit stronger migrations, and this has been
observed for Daphnia hyalina galeata
and Daphnia galeata mendotae,
respectively (Guisande et al., 1991; Lampert, 1992)
Both cyclopoid and diaptomid copepodites, as well as mature Leptodiaptomus
minutus, also showed a significant relationship between body length
and depth of migration. Wells
(1960) found that diaptomus copepodites in Lake Michigan did not migrate
as strongly as the adults. For
the calanoid copepod Limnocalanus
macrurus, both Carter (1969) and Wilson and Roff (1973) found that
later copepodite instars occurred deeper in the water column than earlier
instars.
Conclusions
Members
of both the cladocera and the copepoda were found to
undertake vertical migrations below 20 m. These migrations were largely confined to Lakes Michigan and Huron,
although the dominant cladoceran in eastern Lake Erie, Bosmina
longirostris, was found to migrate almost entirely out of the top 20 m
during the day.Therefore, care should be taken when interpreting data from these
lakes at sites sampled during the day with 20 m tows. Given the plasticity of migratory behavior documented in the
literature, even data from lakes in which migration was not documented in
the present study should be approached with caution, since differences in
predator community structure could have induced migration in the past in
species not presently exhibiting such behavior. Underestimates of abundances resulting from 20 m tows are likely to
be exaggerated when abundances are converted to biovolumes, particularly
for cladocerans, by the tendency of larger animals to undertake deeper
migrations. Previous findings of spatial patterns in zooplankton data generated
from 20 m tows should be approached with extreme caution. In cases where the time of sample collection is known, a possible
ameliorative measure is to exclude historical data collected during the
day from any future trends analysis. However, the large copepods Limnocalanus
macrurus, Senecella calanoides and Leptodiaptomus sicilis have depth preferences substantially below 20 m, and have
most likely always been underestimated by 20 m tows, regardless of time of
collection.
|
