1370
Limnol. Oceanogr.,
44(6), 1999, 1370–1380
䉷
1999, by the American Society of Limnology and Oceanography, Inc.
Mesozooplankton influences on the microbial food web: Direct and indirect trophic
interactions in the oligotrophic open ocean
Albert Calbet and Michael R. Landry
Department of Oceanography, University of Hawaii at Manoa, 1000 Pope Road, Honolulu, Hawaii 96822
Abstract
The phytoplankton in warm oligotrophic regions of the open oceans is dominated by
⬍
2-
m cells too small for
efficient direct consumption by mesozooplankton. However, these primary producers are hypothetically linked to
higher trophic levels via the cascading impacts of mesozooplankton grazing on intermediate consumers. To assess
the magnitudes of these indirect trophic linkages, grazing experiments, involving different concentration treatments
of the mixed mesozooplankton community, were performed during cruises in the subtropical North Pacific at station
ALOHA. Mesozooplankton fed on a diverse assemblage of microzooplankton and nanoheterotrophs
⬎
5
m, and
their predation indirectly enhanced net growth rates of phytoplankton and 2–5-
m heterotrophs. Increasing the
concentration of mesozooplankton also enhanced growth rates of heterotrophic bacteria, but this was more likely
the result of organic enrichment than trophic transfer. Scaled to their natural abundance, the indirect grazing impacts
of mesozooplankton on lower trophic levels are small, accounting for
⬍
0.005 d
⫺1
of the growth rates of each prey
category examined. Thus, the larger consumers appear to exert little net influence on the dynamics at the base of
the food web. In contrast, size-fraction manipulations of consumers between 2 and 20
m (i.e., the nanozooplankton)
elicited strong responses among bacterial populations indicative of tightly coupled predatory chain of at least two
steps. Given the present results, detailed studies of the interactions among pico- and nanoplankton appear to be the
most profitable avenue for improving our understanding of community structure and function in this region and for
acquiring useful data for developing and validating ecosystem models of the open oceans.
Warm oligotrophic waters of the subtropical North Pacific
support a complex planktonic community with pico-sized
(0.2–2
m) phytoplankton providing most of the primary
production and combined auto- and heterotrophic bacteria
dominating community biomass (Campbell and Vaulot 1993;
Letelier et al. 1993; Campbell et al. 1994, 1997). Such or-
ganisms are largely unavailable to direct utilization by the
crustacean-dominated mesozooplankton because of size con-
straints on feeding mechanisms (Rassoulzadegan and Etien-
ne 1981; Conover 1982; Berggreen et al. 1988; Hansen et
al. 1994). Nonetheless, they are linked in principle to higher
order animals by the cascading influences of mesozooplank-
ton grazing on consumers of intermediate size (Sherr et al.
1986; Sherr and Sherr 1988; Wikner and Hagstro¨m 1988).
The strong coupling of nonadjacent trophic levels via in-
direct cascade effects is a well-established phenomenon in
freshwater systems, where the presence or absence of pi-
scivorous fish can profoundly influence zooplankton com-
munity composition, phytoplankton abundance, water clari-
ty, heat budgets, and mixing depth (Carpenter et al. 1985;
McQueen et al. 1986; Kerfoot 1987; Mazumder et al. 1990;
Acknowledgments
We are grateful to the captain and the crew of the R/V Moana
Wave for their help in making the work possible even in rough
weather conditions. We also thank S. Christensen and S. Nunnery
for their assistance with the experiments, H. Nolla for his patience
in the teaching of flow cytometric techniques, and J. Constantinou
for his logistical support and comments on the first version of the
manuscript. This work was supported by National Science Foun-
dation grant OCE-9218152 to M.R.L. and a postdoctoral fellowship
(EX96-46651369) from the Ministerio de Educacio´n y Cultura
(Spain) to A.C. Contributions 4772 from the School of Ocean and
Earth Science and Technology, University of Hawaii at Manoa and
504 from the U.S. JGOFS Program.
Pace et al. 1998). In contrast, this area has received little
attention in marine systems, despite recent evidence that the
harvesting of the oceans is systematically diminishing the
average trophic level of top consumers and concerns about
long-term shifts in community structure (Dayton et al. 1995;
Myers et al. 1997; Pauly et al. 1998).
The subtropical open oceans represent the oligotrophic ex-
treme in the spectrum of marine ecosystems and are conse-
quently an unlikely place to expect cascade effects related
to fisheries. Nonetheless, such systems are vastly important
on a global scale and generally the least studied of all marine
systems with regard to community trophic structure and
function. Moreover, because of the presence of U.S. Joint
Global Ocean Flux Study (JGOFS) time-series monitoring
programs in the subtropical Atlantic and Pacific Oceans,
these regions have become an important focus of efforts to
develop models of open-ocean ecosystems (e.g., Doney et
al. 1996; Hurtt and Armstrong 1996; Lawson et al. 1996).
The present study examines the hypothetical coupling among
food web components from bacteria to mesozooplankton in
the subtropical north Pacific. Our specific goal was to quan-
tify direct and indirect influences of mesozooplankton on
lower levels of the food web. In addition, we sought to begin
to establish a deeper understanding of trophic interactions
that would be relevant to modeling data needs for this re-
gion.
Materials and methods
Study sites—The present experimental work was con-
ducted on shipboard during spring and summer cruises of
the JGOFS Hawaii Ocean Time-series (HOT) Program. The
main sampling site was Stn. ALOHA (22.75
⬚
N, 158
⬚
W), ap-
proximately 100 km north of the island of Oahu, Hawaii
1371Mesozooplankton and the microbial food web
Table 1. Initial conditions for zooplankton grazing experiments. Water was collected from the surface (2 m) or deep chlorophyll
maximum (105–120 m), and experiments focused on grazing of the MESO (200–2,000-
m) or MICRO (60–200-
m) size fractions.
HBACT, 2–5 HNAN, and
⬎
5 HNAN denote initial abundances (cells/ml) for heterotrophic bacteria, 2–5-
m heterotrophs, and
⬎
5-
m
heterotrophs, respectively. Chlorophyll estimates are given as concentrations (
g/L) of the
⬍
2-
m size fraction (
⬍
2 Chl a), the community
total (Tot Chl a), and the % of total
⬍
2
m. Values in parentheses are standard errors of mean estimates.
Date Depth (m) Grazers HBACT 2–5 HNAN
⬎
5 HNAN
⬍
2 Chl a Tot Chl a %
⬍
2 Chl
4 June 120 MESO
⫹
MICRO 3.0
⫻
10
5
(0.4
⫻
10
5
)
110
(4)
23
(1)
0.169
(0.025)
0.177
(0.022)
95.5
26 June 105 MESO 1.6
⫻
10
5
(0.2
⫻
10
5
)
51
(8)
20
(1)
0.290
(0.021)
0.297
(0.019)
97.6
8 July 2 MESO 5.7
⫻
10
5
(0.5
⫻
10
5
)
108
(15)
25
(4)
0.042
(0.001)
0.071
(0.001)
59.1
9 July 115 MESO 4.6
⫻
10
5
(0.1
⫻
10
5
)
55
(8)
17
(1)
0.169
(0.016)
0.215
(0.010)
78.6
10 July 120 MICRO 4.4
⫻
10
5
(0.6
⫻
10
5
)
91
(14)
15
(4)
0.154
(0.001)
0.166
(0.010)
92.7
1 August 2 MESO 7.8
⫻
10
5
(0.1
⫻
10
5
)
221
(47)
31
(6)
0.034
(0.001)
0.052
(0.030)
65.4
(Karl and Lukas 1996). However, one experiment (8 July,
Table 1) used water collected at the Kahe Point test station
(21.48
⬚
N, 158.37
⬚
W; 15 km west of Oahu). These locations
have been sampled approximately monthly since 1988 and
are considered representative of the oligotrophic subtropical
Pacific with regard to the abundance, biomass, and produc-
tion of the plankton community (Campbell and Vaulot 1993;
Letelier et al. 1993; Campbell et al. 1997).
Microbial food web interactions—Experimental incuba-
tions with prescreened seawater were used to identify the
size fractions of consumers most responsible for direct graz-
ing impact on picoplankton populations. Water from the
deep chlorophyll maximum (DCM) at 110 m was prefiltered
through polycarbonate membrane filters of varying pore size
(1, 2, 5, 8, and 20
m) and placed in 250-ml polystyrene
culture flasks (Corning). Four replicate flasks were prepared
for each size-fractionated treatment, and an additional four
flasks were run with unfiltered natural seawater (controls).
The flasks were incubated for 24 h in a seawater-cooled in-
cubator screened with neutral density fabric to 1% of surface
light intensity. Before and after the incubation, 2-ml subsam-
ples were preserved with paraformaldehyde (0.2% final con-
centration), frozen in liquid nitrogen, and stored at
⫺
85
⬚
C
(Vaulot et al. 1989). These subsamples were subsequently
thawed and stained with Hoechst 33342 (1
g ml
⫺1
) (Mon-
ger and Landry 1993), and population abundances of het-
erotrophic bacteria and Prochlorococcus sp. were enumer-
ated using a Coulter EPICS 753 flow cytometer equipped
with two 5-W argon lasers, an MSDS volume-control sam-
pling, and Cytomation CICERO software (Campbell et al.
1997). Net growth rates of the bacterial populations in the
different treatments were computed from the ratios of initial
and final abundance estimates assuming exponential rates of
increase or decrease during the incubation.
Direct feeding interactions of metazooplankton—Three
experiments were conducted on 4 June 1997 to assess the
direct feeding impacts of net-collected meso- and microzoo-
plankton on phytoplankton and the predatory preferences of
mesozooplankton feeding on microzooplankton. Mesozoo-
plankton (MESO
⫽
animals in the 200–2,000-
m size frac-
tion) were collected using a 1-m
2
plankton net with 200-
m
mesh and 7-liter bag as a nonfiltering cod end. Microzoo-
plankton (MICRO
⫽
net-collected zooplankton in the 60–
200-
m size fraction; in this study dominated by copepod
nauplii and codepodites) were collected using a 0.5-m-di-
ameter plankton net with 60-
m mesh. Both nets were towed
obliquely from 100 m to the surface at about 1 knot. Once
on deck, the contents of the cod-end bags were carefully
poured into a 10-liter insulated container, and experimental
organisms were gently sorted by reverse flow through 15-
cm submerged sieves of 200 or 2,000
m mesh for MICRO
and MESO, respectively. Only free-swimming and healthy
organisms that passed through the respective sieves were
used in the experiments.
For the grazing experiments on phytoplankton, aliquots of
each zooplankton size fraction (0.9 mg dry weight for MI-
CRO and 4.8 mg dry weight for MESO) were placed into
2.3-liter polycarbonate bottles filled with water from the
deep chlorophyll maximum (DCM) collected from 110 to
120 m depth with Niskin bottles on a CTD rosette. This
water was filtered through 60- and 200-
m mesh for the
MICRO and MESO experiments to remove ambient grazers
in the respective size fractions. Nitrate (4.5
M NaNO
3
) was
added to all bottles to override differences in zooplankton
excretion among treatments. Four replicates and three con-
trols without zooplankton were prepared per treatment. An
additional group of three bottles was filled with water filtered
through a 2-
m polycarbonate membrane filter to assess
phytoplankton growth without grazers larger than 2
m. The
incubations began at night and were run for 14 h in a sea-
water-cooled incubator screened to 1% of surface irradiance.
The bottles were gently mixed one or two times during the
incubations to reduce settling. Subsamples of the water were
taken for chlorophyll analyses in triplicate at the start of the
1372 Calbet and Landry
Fig. 1. Effects of removal of different size classes of protistan
consumers on net growth rates of heterotrophic bacteria (HBACT)
and Prochlorococcus in the subtropical north Pacific. Treatments
are seawater samples filtered through polycarbonate membrane fil-
ters of 1–20-
m pore size prior to incubation; control
⫽
unfiltered
sample. Vertical bars show standard errors of four replicates.
experiments and from each bottle at the end. For each sub-
sample, an aliquot of 150 ml was filtered thorough a GF/F
filter (total chlorophyll a [Chl a]), and a similar aliquot was
prefiltered through a 2-
m polycarbonate membrane before
GF/F filtration (
⬍
2
m Chl a). The filters were initially fro-
zen in liquid nitrogen and later extracted for 48 h in 90%
acetone and analyzed on a Turner fluorometer by standard
techniques (Strickland and Parsons 1972). Net instantaneous
growth rates of the two chlorophyll fractions were estimated
from initial and final concentrations assuming exponential
rates of change during the incubations.
Incubations to assess the predatory pressure and possible
prey selection of MESO feeding on MICRO were prepared
in 2.3-liter bottles filled with surface seawater and aliquots
of 4.9 mg dry weight of MESO. The treatments consisted
of four different concentrations of MICRO, with three rep-
licates and two controls without MESO for each treatment.
Following the 14-h incubations, the organisms were retained
on a 40-
m submerged sieve and fixed with formalin (5%
final concentration). The retained animals were enumerated
in the laboratory using a dissecting microscope, and inges-
tion rates were calculated according the equations of Frost
(1972) as modified by Landry (1978). Interpretation of these
experiments was simplified by negligible changes in MICRO
abundances in control bottles.
Cascading effects of zooplankton grazing—To quantify
the effects of zooplankton grazing on lower levels of the
food web, we measured the net growth rate responses of
identifiable auto- and heterotrophic populations to increasing
concentration of grazers (Lehman 1980). It was expected
that the relationships between net growth rates of given size
fractions and grazer concentration would alternate between
negative and positive trends for successively lower trophic
levels.
The experimental set-up was essentially similar to that
described for direct grazing effects. Polycarbonate bottles
(2.3-liter) were filled with either DCM (26 June, 9–10 July)
or surface (8 July, 1 August) water amended with a nutrient
mixture (4.5
M NaNO
3
, 0.5
M NH
4
Cl, and 0.2
M
NaH
2
PO
4
) to compensate for differential effects of zooplank-
ton excretion in the treatments. Aliquots of MESO or MI-
CRO (10 July only) were added to two replicate bottles per
treatment and up to five treatments per experiment; in ad-
dition, three control bottles were prepared without animals.
Comparable aliquots of MESO or MICRO were frozen in
liquid nitrogen at the beginning of each experiment for later
assessment of dry weight of the added zooplankton.
The incubations were conducted as previously described
(1% light, 26
⫾
0.5
⬚
C) and lasted for 12–18 h, always in-
cluding the nighttime. At the beginnings and ends of the
incubations, the bottles were subsampled for Chl a, hetero-
trophic bacteria (HBACT), and nanoheterotrophs (HNAN).
Chl a and HBACT samples were taken and analyzed as de-
scribed above. For heterotrophic nanoplankton, 250-ml sam-
ples were preserved with 125
l of alkaline Lugol’s fixative
(0.05% final concentration), followed immediately with the
addition of 5 ml of borate-buffered formalin (2% final con-
centration) and 250
l of sodium thiosulfate (0.3% final con-
centration) to clear the iodine color (Sherr and Sherr 1993).
After fixation, the samples were momentarily stained with
proflavin (12
g/ml final concentration), then stained with
DAPI (8
g ml
⫺1
final concentration) for a few seconds and
filtered onto 2-
m pore-size black polycarbonate membrane
filters (3
m for 8–9 July, 1 August). The filters were mount-
ed in immersion oil (type B) on slides and kept frozen until
analysis with a color image-analysis system consisting of a
Zeiss epifluorescence microscope with a ZVS 3-chip charge-
coupled device video camera connected to a computer. The
images were analyzed using a Zeiss Image Pro Plus software
to facilitate counting and sizing of all heterotrophic and au-
totrophic organisms. About 300 individuals were enumerated
and sized for each sample. We present the results in two size
fractions: heterotrophs between 2 and 5
m (hereafter called
2–5
m HNAN) and heterotrophs larger than 5
m (
⬎
5
m
HNAN). Autotrophic organisms in the larger size fraction
were not sufficiently abundant for meaningful statistical an-
alyses; therefore, these organisms are included more accu-
rately in the Chl a size fractions. The results (net growth
rates vs. MESO or MICRO biomass) were fit to linear re-
gressions (Lehman 1980). Where necessary, ANCOVA tests
were performed to detect significant differences between re-
gression lines.
Results
Grazing on bacteria and bacterivores—Food web link-
ages in the microbial component of the plankton community
are illustrated by the net growth rates of heterotrophic and
phototrophic (Prochlorococcus sp.) bacteria in the various
size-fractionated treatments (Fig. 1). Removal of the
⬎
20-
m consumers resulted in only a slight decline in bacterial
1373Mesozooplankton and the microbial food web
Fig. 2. Impacts of micro- and mesozooplankton grazers on net
growth rates of total and
⬍
2-
m Chl a in the subtropical north
Pacific. Treatments are
⬍
2-
m filtered water,
⬍
60-
m filtered wa-
ter with and without 0.4 mg dry wt. MICRO L
⫺1
, and
⬍
200-
m
filtered water with and without 2.1 mg dry wt MESO L
⫺1
. Vertical
bars show standard errors of three (without MESO or MICRO) or
four (with MESO or MICRO) replicates.
net growth, suggesting that ambient concentrations of mi-
crozooplankton had little direct impact on lower trophic lev-
els or that their effects are broadly distributed among organ-
isms of different size. In contrast, removal of organisms in
the 5–20-
m size range resulted in substantial decreases in
net growth rates of bacteria. The decrease was particularly
spectacular for Prochlorococcus (approximately
⫺
0.6 d
⫺1
)
and less so for HBACT that dominate biomass and presum-
ably grow much slower on average than autotrophic cells in
the open oligotrophic ocean (Fuhrman et al. 1989; Cho and
Azam 1990).
In comparison to the effects of the 5–20-
m fraction, suc-
cessive removal of organisms
⬎
2
m and
⬎
1
m increased
net bacterial growth rates, restoring them to only slightly less
than they were in unfiltered controls (Fig. 1). Most of this
positive net effect on bacterial growth was achieved follow-
ing removal of the 2–5-
m fraction. Such results suggest
that bacteria are principally the prey of consumers in the 2–
5-
m fraction (2–5
m HNAN). In turn, these ‘‘bacteri-
vores’’ appear to be grazed by
⬎
5-
m nanoplankton (
⬎
5
HNAN). The breakdown of the community into
⬍
2-
m, 2–
5-
m, and
⬎
5-
m fractions is carried over to meso- and
microzooplankton experiments discussed below.
Composition of the microbial plankton community—Table
1 summarizes the size structure of the microbial community
at the study site during June–August 1997. HBACT varied
from 1.6 to 7.8 10
5
cells ml
⫺1
, being typically more abundant
in surface waters than in the DCM. This trend was also
evident for other components of the heterotrophic commu-
nity but opposite for Chl a, presumably due to photoadap-
tation of cellular pigment content. Heterotrophs in the 2–5-
m size range varied from 50 to 220 cells ml
⫺1
, exceeding
⬎
5
m HNAN on average by a factor of 4.7 (range 2.6–
7.1). Total Chl a varied from 0.05 to 0.3
g L
⫺1
, and the
⬍
2-
m fraction represented 91% of the total on average for
experiments involving water from the DCM and 62% of the
total for surface water.
Initial grazing experiments—Initial grazing experiments
with MICRO and MESO showed marked contrasts in chlo-
rophyll responses among treatments (Fig. 2). In the 2-
m
prescreened incubations, chlorophyll net growth was positive
(
艐
0.3 d
⫺1
) and comparable to the Prochlorococcus growth
noted previously when all grazers were removed (Fig. 1).
Prescreening of the water with either 200- or 60-
m mesh
removed the larger size categories of grazers, MESO or
MESO
⫹
MICRO, respectively, leaving the microbial portion
of the food web, including most heterotrophic protists, intact.
Under these circumstances and the low light levels of the
experimental incubations, net growth rates of
⬍
2-
m pico-
phytoplankton were strongly negative, about
⫺
0.5 d
⫺1
. The
addition of either MESO or MICRO grazers to treatment
bottles containing the protistan assemblage enhanced the net
growth rates of picophytoplankton relative to controls, al-
though the rates were still negative and substantially less
than those when all grazers
⬎
2
m were removed.
While MESO and MICRO showed comparable effects on
the
⬍
2-
m chlorophyll fraction, their impacts on total chlo-
rophyll were different (Fig. 2). Total Chl a grew at a low,
but net positive rate in water prescreened to remove both
MICRO and MESO fractions, and the addition of MICRO
resulted in a slightly negative net rate. In comparison, net
growth of total Chl a in the absence of MESO was more
negative than with added MESO.
Ingestion rates of MESO feeding on MICRO are shown
in Fig. 3. Only data for copepod nauplii and copepodites are
presented because other organisms, like ciliates or dinofla-
gellates, were too rare for statistically reliable enumeration.
MESO showed a clear feeding preference for copepodites
over nauplii. Due to the large difference in mean sizes of
these potential prey organisms, the twofold predatory pref-
erence for copepodites is substantially greater in terms of
biomass consumed.
The trophic cascade—The indirect food web effects im-
plied in the 4 June experiments (Fig. 2) were investigated
on subsequent cruises over a range of grazer concentrations
(Figs. 4, 5). In the experiments conducted in this manner,
the slopes of individual regression analyses of net prey
1374 Calbet and Landry
Fig. 3. Functional relationships between ingestion rates of me-
sozooplankton (prey consumed mg dry wt
⫺1
h
⫺1
) and abundances
of nauplii (open symbols) and copepodites (dark symbols) in the
subtropical north Pacific.
growth rate versus increasing grazer biomass were frequent-
ly not significantly different from zero. However, the slopes
were consistently positive or negative for individual prey
categories, and the mean slope for each prey type was sig-
nificant at either P
⬍
0.05 or P
⬍
0.01 (Table 2).
Among pigmented cells, both the
⬍
2-
m Chl a size frac-
tion and abundances of Prochlorococcus showed weakly
positive (means
⫽
0.05 and 0.02 L mg dry wt
⫺1
d
⫺1
, re-
spectively) and generally insignificant trends with increasing
MESO or MICRO biomass (Table 2). The
⬎
2-
m Chl a
fraction responded more dramatically (mean
⫽
0.40 L mg
dry wt
⫺1
d
⫺1
). The projected net growth rates in the absence
of large grazers (i.e., the Y-intercepts of the regression equa-
tions) were variable but always positive for larger phyto-
plankton. For the
⬍
2-
m Chl a fraction, net growth rates
ranged between
⫺
0.4 and
⫹
0.5 d
⫺1
, being positive for ex-
periments conducted with surface water and negative with
water from the DCM. These consistent differences in growth
rate intercepts (Table 3) most likely reflect photoadaptive
effects, as the dominant pigmented organism in this size cat-
egory, Prochlorococcus, showed relatively low net rates of
change in cell abundance (
⫺
0.1 to 0.15 d
⫺1
) during the in-
cubations and no difference between deep and shallow sam-
ples (Table 2). The main difference in the responses of lower
trophic levels to variations in biomass of MESO versus MI-
CRO grazers was in the
⬎
2-
m Chl a. The net growth rate
intercept was higher when MESO was varied, but the net
growth enhancement effect was almost threefold higher
(0.89 versus 0.30 L mg dry wt
⫺1
d
⫺1
) when MICRO biomass
was increased (Tables 2, 3).
Among heterotrophic populations, net growth rates of
HBACT and 2–5-
m HNAN were enhanced by increasing
concentrations of MESO or MICRO (means
⫽
0.26 and 0.49
L mg dry wt
⫺1
d
⫺1
, respectively), but
⬎
5-
m HNAN de-
creased (mean
⫽ ⫺
0.23 L mg dry wt
⫺
1
d
⫺
1
) (Fig. 5, Table
2). No significant differences were found between MESO
and MICRO experiments except for the HBACT intercept
(Table 3). The regression slopes for HBACT and 2–5-
m
HNAN were similar for experiments conducted with DCM
water (26 June, 9 and 10 July). However, the net growth
responses of 2–5-
m HNAN exceeded HBACT by 3–11-
fold in experiments with surface water (8 July, 1 August).
Discussion
Food web organization—Among the clearest results of the
present study were the effects of removing various size frac-
tions of heterotrophic nanoplankton on the net growth rates
of picoplankton (Fig. 1). Removing consumers between 5
and 20
m had a significant negative effect on the net
growth of HBACT and particularly Prochlorococcus, while
removing the 2–5-
m HNAN caused the net rates to return
to control levels. These results suggest strong trophic link-
ages at the base of the food web—from picoplankton to
⬍
5-
m HNAN that consume bacterial-sized food, to 5–20-
m
HNAN that keep bacterivores in check.
Within the micro- and mesozooplankton size categories,
the food web linkages become more diffuse and unstruc-
tured. For example, both net-collected MICRO and MESO
exerted a direct grazing impact (i.e., negative influence)
upon
⬎
5-
m HNAN (Fig. 5). In addition, MESO showed a
strong preference for preying on larger MICRO (Fig. 3). The
simplest scenario consistent with these results is for MICRO
to feed mainly on the larger HNAN, while MESO feed
broadly on prey in the 5–200-
m range. Protistan micro-
zooplankton (e.g., ciliates) occur in these waters, but their
abundances are typically well below 1 cell ml
⫺1
and their
biomass is vastly outweighed by smaller flagellates (Beers
et al. 1975). If the grazing role of larger protists was sub-
stantial, we would have expected to see a larger response
when
⬎
20-
m cells were removed by size fractionation
(Fig. 1). Moreover, if protists occupied a significant inter-
mediate trophic link between net-collected MICRO and larg-
er HNAN, increasing concentrations of MICRO grazers
should have resulted in an increase in
⬎
5-
m HNAN. Be-
cause neither of these effects was observed, we conclude
that the protistan grazing pathway is reasonably well repre-
sented by two levels within the nanoplankton size category.
The trophic structure elucidated by the present experi-
ments differs from the fixed 10-fold ratio of predator:prey
sizes as originally proposed for the microbial loop by Azam
et al. (1983). If the predator:prey size ratio was a factor of
10 by linear dimension, HBACT (mean equivalent spherical
diameter [ESD]
艐
0.5
m) and Prochlorococcus (mean ESD
艐
0.7
m) would be consumed by HNAN in the 5–7-
m
range, and they in turn would be preyed upon by protistan
microplankton of 50–70
m size. The trophic structure sug-
gested by Fig. 1 is in better agreement with a predator:prey
size ratio on the order of 3 to 4 for flagellates, as summarized
by Hansen et al. (1994). An average ratio of 4 would yield
bacterivores of about 2.4
m and secondary consumers of
about 10
m. In experiments in which predation rates on
bacterial-sized particles were measured directly in size-frac-
