christian and lucz
Ecological Modelling 117 (1999) 99 – 124
Organizing and understanding a winter’s seagrass foodweb
network through effective trophic levels
Robert R. Christian *, Joseph J. Luczkovich
Biology Department, East Carolina Uni6ersity, Green6ille, NC 27858-4353, USA
Received 9 June 1998; accepted 5 January 1999
Abstract
Trophic structure of ecosystems is a unifying concept in ecology; however, the quantification of trophic level of
individual components has not received the attention one might expect. Ecosystem network analysis provides a
format to make several assessments of trophic structure of communities, including the effective trophic level (i.e.
non-integer) of these components. We applied network analysis to a Halodule wrightii community in Goose Creek
Bay, St. Marks National Wildlife Refuge, Florida, USA, during January and February 1994 where we sampled a wide
variety of taxa. Unlike most applications of network analysis, the field sampling design was specific for network
construction. From these data and literature values, we constructed and analyzed one of the most complex, highly
articulated and site specific foodweb networks to be done. Care was taken to structure the network to reflect best the
field data and ecology of populations within the requirements of analysis software. This involved establishing
internally consistent rules of data manipulation and compartment aggregation. Special attention was paid to the
microbial components of the food web. Consumer compartments comprised effective trophic levels from 2.0
(herbivore/detritivore) to 4.32 (where a level 4.0 represents ‘secondary carnivory’), and these values were used to
organize data interpretation. The effective trophic levels of consumers tended to aggregate near integer values, but the
spread from integer values increased with increasing level. Detritus and benthic microalgae acted as important sources
of food in the extended diets of many consumers. ‘Bottom-up’ control appeared important through mixed trophic
impact analysis, and the extent of positive impacts decreased with increasing trophic level. ‘Top-down’ control was
limited to a few consumers with relatively large production or biomass relative to their trophic position. Overall,
ordering results from various network analysis algorithms by effective trophic level proved useful in highlighting the
potential influence of different taxa to trophodynamics. Although the calculation of effective trophic level has been
available for some time, its application to the evaluation of other analyses has previously not received due
consideration. © 1999 Elsevier Science B.V. All rights reserved.
Keywords: Seagrass community; Network analysis; Effective trophic level; Carbon flow
* Corresponding author. Tel.: +1-252-3281835; fax: + 1-252-3284178.
E-mail address: christianr@mail.ecu.edu (R.R. Christian)
0304-3800/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 3 8 0 0 ( 9 9 ) 0 0 0 2 2 - 8
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
100
1. Introduction The seagrass communities along the coasts of
the southeastern United States and the Gulf of
Trophic structure is one of the primary ways by Mexico support substantial populations of ben-
which ecologists organize communities and thos, nekton and waterfowl (Zieman and Zieman,
ecosystems. Although trophic levels are often con- 1989). The primary producers in these communi-
sidered as discrete integers (Lindeman, 1942); in- ties may include several species of Submerged
dividual consumers, their populations or guilds Aquatic Vegetation (SAV), their epiphytes, phyto-
often feed across several trophic levels (Odum and plankton, benthic microalgae, and macroalgae.
Heald, 1975). Thus, populations or guilds may Birds and large fish represent important top con-
have ‘effective trophic levels’ that are fractional sumers, and the relative importance of each or-
(Odum and Heald, 1975; Levine, 1980). For ex- ganism may vary with season. In winter waterfowl
ample, a consumer that feeds as a herbivore (level are particularly abundant. The links between the
2) for 50% of its diet and as a primary carnivore primary producers and the top consumers are
(level 3) for 50% would have an effective trophic often poorly understood, with several trophic
level of 2.50. The calculations are done by both of steps between producer and top consumers. These
the most commonly used software packages for trophic steps may be mediated by microbes and
ecosystem network analysis; ECOPATH II (Chris- animals in both sediments and water column.
tensen and Pauly, 1992) and NETWRK4 (Ulanow- Ecosystem network analysis has been used to
icz, 1987). assess the foodweb interactions (Wulff et al.,
Although effective trophic level has been used 1989; Christensen and Pauly, 1993). The links
to characterize food webs, the full application of between primary producers and birds, a poten-
this classification has not been realized. Odum tially important consumer group, is rarely in-
and Heald (1975) used it to group various taxa cluded in complex network analyses (Baird and
into common feeding categories. Other re- Ulanowicz, 1993; Biujse et al., 1993). An effort
searchers have used it to compare trophic struc- was made here to include these consumers and
tures among ecosystems (e.g. Ulanowicz, 1984; evaluate their potential roles in trophodynamics.
Ulanowicz and Wulff, 1991). Recently, Pauly et The foundation for our research has been a
al. (1998) applied the concept to evaluate fishery priori collection of data to support the construc-
trends. It has even been used to examine theoreti- tion and analysis of a winter’s seagrass foodweb.
cal issues of energy flow (Burns, 1989). The em- Sampling was specifically designed for network
phasis in all of these studies has been at the construction and to be inclusive of the full range
ecosystem level. Little effort has been expended of trophic groupings (Luczkovich et al., 1997;
on the actual interaction of specific components submitted). We measured standing stocks of mi-
and their contribution to within-system regula- crobes, benthos, plankton, nekton, birds and or-
tion. Effective trophic level can be used as a ganic carbon as well as selected flows and diets.
scaling metric for other analyses to infer various The collection was a joint effort by us and staff of
attributes and contributions to trophodynamics. the National Wetlands Research Center, US Geo-
For example, populations with higher effective logical Survey (USGS). The focal ecosystem was
trophic levels would be expected to contribute less the seagrass communities within Goose Creek Bay
to the energetics of the ecosystem than those with of St. Marks National Wildlife Refuge, St. Marks,
lower levels. Deviations from this trend may indi- FL, USA. Further, Livingston and coworkers
cate that a consumer is particularly important or have amassed considerable information on the
unimportant to the food web. Also, the potential ecology of the northern Gulf of Mexico and its
for top-down or bottom-up control may be re- coastal and estuarine ecosystems (Heck, 1979;
lated to a population’s effective trophic level. In Stoner, 1979, 1980; Livingston, 1980, 1982, 1984;
this report effective trophic level was used to Lewis and Stoner, 1981; Leber, 1983; Lewis, 1984;
organize a seagrass food web and investigate these Luczkovich, 1987). From the field and laboratory
issues in that context. studies and the literature, foodweb networks were
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 101
constructed and analyzed with ECOPATH II for trophic position and potential importance of the
winter 1994. Three broad objectives were iden- different taxa in relation to position. Further-
tified with corresponding manuscripts. more, the issue of adapting field data for network
Luczkovich et al. (1997; submitted) addressed analysis is addressed.
sampling design related to the needs for network
construction following the guidelines of Cohen et
al. (1993). In Baird et al. (1998) networks were 2. Methods
reconstructed for analysis by NETWRK4 and evalu-
2.1. Sample site and design
ated for uncertainties of input and output vari-
ables with emphasis on systems-level attributes. In
the present paper the trophic structure of the Sampling was conducted from January and
system is evaluated with special attention to February 1994. Three sites were sampled in each
Fig. 1. Sample sites within St. Marks National Wildlife Refuge, St. Marks, Florida, USA. Numbered areas are sample sites.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
102
month: in January sites 1, 2, and 3; in February Zooplankton samples were obtained with a 90
mm mesh plankton net and preserved. In the
sites 1, 2, and 4 (Fig. 1). Sites 1 (Live Oak Island)
and 2 (Wakulla Beach) were within Goose Creek laboratory, the samples were sieved through a
Bay and similar in community structure and hy- series of screens, and the contents of each screen
drologic regime (Baird et al., 1998). Therefore, were counted as various taxa. Sieve fractions were
then dried at 60°C for 48 h and the biomass l − 1
only they were used for this network construction
and analysis. At each site sampling occurred at 3 in the original sample calculated, using the counts
transects, running perpendicular to the shore and to estimate proportional contribution of each tax-
extending through a Halodule wrightii community onomic group.
to approximately 150 m offshore. As a general Meiofauna were sampled by coring sediments,
minimum one sample was collected from each of preserved, and later separated from sediment,
the three transects within a site, but the different sorted, identified to the lowest possible taxon, and
variables measured required different sampling enumerated. Dimensions of representative organ-
procedures. Some variables were collected with isms were measured for conversion to biomass
much greater replication. (Higgins and Thiel, 1988).
To estimate the standing stock of macroinverte-
2.2. Sampling and sample analysis o6er6iew brates associated with seagrasses, 30 cores (7.62-
cm inside diameter) were taken per site (Lewis
Methods for field sampling and diet determina- and Stoner, 1981). The core samples were sieved
through 500 mm mesh in the field and placed into
tions are described in depth elsewhere by
Luczkovich et al. (1997, submitted) and Baird et jars with 10% formalin with rose bengal stain. In
al. (1998). Here we give an overview. the laboratory, animals in the sieved portions
Primary productivity and standing stocks of were sorted and identified to taxonomic groups.
primary producers were estimated largely by Polychaetes were identified to family level. Am-
USGS personnel directed by W. Rizzo and H. phipods, molluscs, decapod crustaceans and
Neckles. Ground cover along transects at each isopods were identified to species. Other inverte-
site was determined with periodic biomass sam- brate groups were identified as necessary. Biomass
pling. Biomass was divided by macrophyte species of each taxon was determined after drying. In the
(above- and below-ground), microepiphytes, and case of molluscs, ophuroids, polychaetes, isopods,
macrophytic algae. Benthic microalgal biomass decapods, and amphipods, ash-free dry masses
was estimated from chlorophyll a content in sur- were obtained by ashing representative samples
face layers of cores from each site. Phytoplankton and subtracting the mass of the remaining ash
biomass was estimated from aquatic chorophyll a from the dry mass. All ash-free dry masses were
concentrations. Benthic microalgal and phyto- converted to g carbon by multiplying by 0.45; for
plankton productivities were estimated from samples in which dry masses alone were deter-
changes in dissolved oxygen concentrations with mined, they were converted to g carbon by multi-
incubation in light and dark. plying by 0.40 (Jørgensen et al., 1991).
Benthic bacteria and sediment organic matter A technique developed for this study, the bar-
were sampled at each transect by coring to 5 cm, rier seine, and gill nets were used to sample fishes
and water samples were taken at each transect for and large mobile decapods at each station. All
dissolved and particulate carbon, bacterioplank- fishes caught in both gill nets and seines were
ton and planktonic microprotozoans. The densi- preserved in 10% formalin and taken back to the
ties of these organisms were estimated by lab where they were identified, counted, and
epiflourescence microscopy with appropriate weighed.
flourochromes. During each month at Wakulla Waterfowl standing stocks were estimated by
Beach, bacterioplankton growth and grazing rates surveys conducted during field campaigns and by
were estimated by modification of the method of D. Everette (The Florida State University, De-
Landry and Hassett (1982). partment of Biological Science). Everette made
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 103
five trips to Wakulla Beach and Live Oak Island were derived from allometric relationships to
from 4 February to 14 March 1994. On each body mass as summarized by Peters (1983) and
occasion and at each site, he counted birds within lowered to 75% of annual values to correct for
a 500×500 m2 area for 1 h. winter temperatures. QB values were then derived
To determine the structure of the diet matrix assuming set fractions of GE based on diet, where
required for ECOPATH II, dietary analyses were the fractions for detritivores, herbivores, omni-
conducted. Stomach content analysis was per- vores and carnivores were 0.15, 0.20, 0.20, and
formed on the most common fish species found in 0.25, respectively. Homeotherms (birds) were as-
the collections. Stomach contents of the fishes sumed to produce 1.5% of body mass per day
were analyzed following the sieve fractionation with food gross efficiencies of either 3 or 6% of
methodology of Carr and Adams (1972, 1973) as consumption. Lastly, diet distributions were esti-
modified by Luczkovich and Stellwag (1993). In mated for each consumer compartment from ei-
other cases, where fish samples were too small to ther gut analyses of field samples when available
conclude anything about diets, and for the inver- or the literature.
tebrate groups, estimates of dietary composition Ecosystem network analysis is actually a collec-
were obtained from the literature (see Baird et al., tion of mathematical algorithms to evaluate the
1998). structure of networks and ecosystems by infer-
ence. For this presentation interpretive efforts
2.3. Modelling and analysis approach concentrated on evaluating trophic structure and
the impacts of different organisms on trophody-
Biomass, given as mgC m − 2, was estimated for namics. ECOPATH II outputs of effective trophic
the various taxa collected in January and Febru- level (Levine, 1980), the mixed trophic impact
ary from sites 1 and 2. Estimation came from matrix (Ulanowicz and Puccia, 1990) and om-
either direct measurement of dry mass or conver- nivory index (Christensen and Pauly 1992) were
sion from density based on estimated dimensions used here. The algorithms for the three analyses
of the organisms. Taxa were then organized to are found in the cited references. Documentation
represent living compartments based on probable for these and other algorithms within the ECO-
diet and life history characteristics. The ‘detritus’ PATH II software are found in Christensen and
compartment was the sum of sediment organic Pauly (1992).
carbon and dissolved and estimated non-living
particulate carbon in the water column. The con-
version of volumetric to aerial data assumed a 3. Results and discussion
depth of 0.75 m for the water column and 5 cm
3.1. Input 6ariables and applying field data to
for the sediment.
ECOPATH II was used for network analysis network construction
(Christensen and Pauly, 1992). Version 2.1 was
used initially, but studies were completed with A listing of the compartments used in the food-
version 3.0 for Windows. The program required web network of a H. wrightii community in Goose
estimates of biomass (B) per compartment, Pro- Creek Bay averaged from January and February
ductivity: Biomass (PB), Consumption: Biomass 1994 is presented in Table 1. There are 48 com-
(QB), fraction of unassimilated food, and/or some partments. As with most representations of food
combined variable (e.g. Gross food conversion webs, the living compartments represent different
Efficiency (GE as PB/QB)). Some of these values degrees of aggregation (Cohen et al., 1993). Com-
were derived from field data, especially primary partments range from single species (e.g. 25 and
productivity of algae; but most came from litera- 26) to a few species (e.g. 24 and 38) to large
ture. Three important sources were Christensen groupings of taxa, especially of small organisms
and Pauly (1993), Jørgensen et al. (1991) and (e.g. 1 and 3). Similarities in diet and habitat are
Peters (1983). PB values for most poikilotherms the two main distinguishing characteristics for a
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
104
Table 1
networka
Summary of compartments for ECOPATH II
No. Compartment name Biomass PB (d) QB (d) Unassimilated Growth Fraction
(mgC/m2) food efficiency imported
1 Benthic bacteria 262.50 0.2500 1.0000 0.10 0.25
2 Microfauna 94.00 0.2000 0.6066 0.20 0.33
3 Meiofauna 1038.50 0.0476 0.1590 0.50 0.30
4 Bacterioplankton 10.90 1.5214 6.0855 0.00 0.25
5 Microprotozoa 4.70 1.0000 3.1250 0.20 0.32
6 Epiphyte-grazing amphipods 69.00 0.0103 0.0513 0.50 0.20
7 Suspension-feeding molluscs 6.76 0.0073 0.0364 0.50 0.19
8 Hermit crabs 178.52 0.0033 0.0222 0.50 0.14
9 Spider crabs (herbivores) 0.07 0.0002 0.0012 0.50 0.15
10 Omnivorous crabs 175.08 0.0007 0.0033 0.50 0.20
11 Blue crabs 12.74 0.0008 0.0031 0.50 0.25 0.1
12 Isopods 61.22 0.0066 0.0328 0.50 0.20
13 Brittle stars 370.83 0.0026 0.0129 0.50 0.20
14 Deposit-feeding peracaridan crustaceans 73.60 0.0086 0.0570 0.50 0.15
15 Herbivorous shrimps 24.58 0.0033 0.0165 0.50 0.20
16 Predatory shrimps 50.66 0.0031 0.0126 0.50 0.25
17 Catfish and stingrays 54.87 0.0025 0.0100 0.20 0.25 0.9
18 Tonguefish 1.44 0.0150 0.0599 0.20 0.25
19 Gulf flounder and needlefish 35.14 0.0061 0.0243 0.20 0.25
20 Southern hake and sea robins 9.34 0.0101 0.0402 0.20 0.25
21 Atlantic silverside and bay anchovies 7.90 0.0105 0.0418 0.20 0.26
22 Sheepshead minnow 8.39 0.0105 0.0700 0.20 0.16
23 Killifishes 2.26 0.0126 0.0628 0.20 0.21
24 Gobies and blennies 1.86 0.0183 0.0733 0.20 0.25
25 Pinfish 2.44 0.0351 0.1402 0.20 0.25
26 Spot 98.31 0.0289 0.1156 0.20 0.25
27 Pipefish and seahorses 1.41 0.0267 0.1066 0.20 0.25
28 Red drum 35.35 0.0026 0.0105 0.20 0.25 0.535
29 Deposit-feeding gastropods 974.93 0.0049 0.0325 0.50 0.15
30 Predatory gastropods 283.36 0.0099 0.0496 0.50 0.20
31 Epiphyte-grazing gastropods 6.46 0.0162 0.0811 0.50 0.20
32 Other gastropods 15.49 0.0110 0.0549 0.50 0.20
33 Deposit-feeding polychaetes 132.10 0.0104 0.0692 0.50 0.15
34 Predatory polychaetes 84.16 0.0043 0.0170 0.50 0.24
35 Suspension-feeding polychaetes 6.74 0.0129 0.0647 0.50 0.20
36 Zooplankton 2.50 0.0660 0.3301 0.50 0.20
37 Benthos-eating birds 1.89 0.0150 0.2400 0.25 0.06 0.02
38 Fish-eating birds 36.93 0.0150 0.2400 0.25 0.06 0.935
39 Fish and crustacean-eating birds 1.17 0.0150 0.2400 0.25 0.06 0.49
40 Gulls 7.17 0.0150 0.2400 0.25 0.06 0.855
41 Raptors 1.85 0.0150 0.2400 0.25 0.06 0.69
42 Herbivorous ducks 0.35 0.0150 0.2400 0.25 0.06 0.11
43 Halodule 4963.00 0.0020 0.0000 0.00
44 Micro-epiphytes 259.90 0.7500 0.0000 0.00
45 Macro-epiphytes 54.10 0.0300 0.0000 0.00
46 Benthic algae 1073.50 0.0997 0.0000 0.00
47 Phytoplankton 71.10 1.5000 0.0000 0.00
48 Detritus 369500
a
Average input values for winter 1994 in Goose Creek Bay, St. Marks National Wildlife Refuge, FL.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 105
compartmental grouping. Details of groupings are tropods. As calculated for input to the network,
described in Luczkovich et al. (1997) and summa- PB and QB values generally were inversely related
rized in Appendix A. Compartments are ordered to body size, being highest in the plankton.
numerically in the sequence with which data were Growth efficiencies for microbes and poikilo-
entered into ECOPATH II. In general, the order is therms ranged from 0.14 to 0.33. The lower effi-
microbes, benthic and epiphytic arthropods and ciency of 0.06 was used for birds.
bivalves, fish, gastropods and polychaetes, Although the populations of organisms in the
zooplankton, birds and at the end primary pro- field fluctuated over time, construction of a steady
ducers and detritus. state network was attempted. Steady state was
Most taxa were found during both months. considered achieved for any prey grouping that
Catfish and stingrays (17), tonguefish (18), gulf had an ecotrophic efficiency, i.e., fraction of pro-
flounder and needlefish (19), Atlantic silversides duction going to predation, harvest and export
and bay anchovies (21), and gobies and blennies (Christensen and Pauly, 1992) of 1 or less. To do
(24) were not found in January. Sheepshead min- this, one has several choices for modification of a
now (22), red drum (28), and killifish (23) were compartment’s attributes; biomass, parameter ra-
not found in February. Herbivorous spider crabs tios associated with metabolism, and food source.
(9) and herbivorous ducks (42) were not found in It was considered that the biomass data were the
January. All of these were included in the winter’s most reliable, as these were collected most di-
network. Values of zero were used for the month rectly. These data were not manipulated to
when the organisms were not present, and the achieve steady state. As described in the Methods
zeros were averaged with values obtained for the section, rules for parameter ratios were internally
month when the organisms were present. consistent for all, or at least related, groupings,
The 48 compartments are associated with 333 and these were not modified. Diet distributions,
individual transformations and transfers: 9 im- especially those from the literature, were subjected
ports, 47 respirations, 230 feeding pathways, and to the greatest manipulation because they were
47 returns to detritus (Tables 1 and 2). Cohen et considered to be a flexible parameter. Diets often
al. (1993) addressed the difficulty in presenting vary significantly across time and space in re-
large food webs through box and arrow diagrams, sponse to availability of different food items
indicating that graphical representations may be (Polis, 1995). If a prey grouping had an
too complicated to be meaningful. Therefore, in- ecotrophic efficiency greater than 1 (i.e. predation
formation used for network construction here is exceeded production in the network), the diet
given in tabular form as required for analysis by distributions of its predators were altered to re-
ECOPATH II. The input variables in Table 1 in- duce predation on it. After all reasonable alter-
clude biomass, PB, QB, their ratio as gross effi- ations of this kind, only three groups were
ciency, fraction of consumed food that is allowed to remain slightly overgrazed; predatory
unassimilated, and fraction of consumption im- shrimp (16), sheepshead minnow (22), and de-
ported from outside the site. The diet matrix in posit-feeding gastropods (29) (Table 3).
Table 2 includes feeding pathways from all food As a first assumption, most organisms were
sources to each consumer, as fractions of the considered to spend their time in the seagrass
consumer’s diet. All consumers contributed to community or in similar communities. Thus, there
detritus through mortality, egestion and excretion. was no import or export of material, unless dic-
As seen in Table 1 the largest biomass was in tated by the organism’s energetic balance and
detritus, primarily because of sediment organic biology. This appeared reasonable for many of
carbon. The primary producers, H. wrightii and the benthos and ichthyoplankton (Tolan et al.,
benthic microalgae had biomasses greater than 1997). During the process of adjusting ecotrophic
103 mgC m − 2. The only consumers to have efficiencies of prey, it became evident that some
biomasses around 103 mgC m − 2 were benthic predators could not be supported by the amount
fauna: i.e. meiofauna and deposit-feeding gas- of prey measured within the system. These preda-
106
Table 2
Diet matrix for winter 1994 in Goose Creek Bay, St. Marks National Wildlife Refuge, FL
Prey number Diet Composition as percentage of total ingestion of predator (designated by c)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
1 65 8 1 9.5 9.5
2 5 4 3 3
3 3 9 10 16.9 1 9 33 43
4 47 14
5 3 6
6 3 8 66 23.1 2 4 30 6
7 0.5 0.5
8 7 74 5
9
10 1 3 0.5 3.5
11
12 20 13
13 8.5
14 1.5 6.6 34 1 13 1 2 17 3
15 5 2.5 10 2.5
16 0.5 27 7
17
18
19
20 0.5
21
22
23
24
25 2 1
26 2 98 41 26 11
27 9
28
29 4.5 18
30
31
32
33 2.6 14.1 1.5 10 5.5
34 0.9 7 4 2
35 1.9 1 0.5
36 1 1 1 3 16
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
37
38
39
40
41
42
43 10 90 25 50 50 1 1
44 30 5 5 50 50 6
45 25 81 56
46 20 29 45 5 37.5 37.5 7 23
47 60 100 40 80 4.9
48 100 10 10 31 45 50 40 50 15 7 16 1
Table 2 (Continued)
Prey number Diet Composition as percentage of total ingestion of predator (designated by c )
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
1 1 0.5 1
2 1 1.5 1
3 60 40 1 18 42.5
4 7
5 5
6 2 27 7 1
7 0.5 0.5
8 5 15 3.5 6
9
10 5 1.5 3
11 0.5
12 12.5
13
14 1.5 12.5 7.5 0.5
15 0.5
16 0.5 0.5 0.5 0.5
17 19
18
19 1 8
20 1 2
21 1 2.5
22 1 0.5 0.8 2 2
23 1 1
24 1
25 1
26 8 5 40.2
27 0.5
28 2
29 30 38 70 23
30
31 0.5
32 0.5 1 1
33 8 2.2 7 5.5 6.5
34 2 0.8 3 3
35 0.5 0.5 0.5
36 0.5 2 1
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
37
38
39
40
41
42
43 62.5
44 100 100 3
45
46 25 2 48 25 99
47 9 88
48 49 69.5 50
107
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
108
tors were either nekton or birds, which are quite ican white pelicans that were frequently found at
able to leave the area within the time scale of the site for much of the winter. As a result of the
minutes to hours. They all had high biomasses importation of carbon associated with the steady
relative to their trophic position and high areal state assumption, the significance of bird feeding
consumption rates (Table 3). Blue crabs (11), on community structure could not be truly
catfish and rays (17), and red drum (28) were the quantified.
nekton groups, with the catfish and rays needing Aggregation of species into trophic guilds is
90% of their diet imported (Table 1). Rays may required for network analysis of most, if not all,
actually eat more frequently within the commu- natural ecosystems. This results from both the
nity, e.g. feeding on polychaetes not readily sam- fact that identification and characterization of all
pled by our techniques (P. Wilbur, personal species in an ecosystem are beyond the abilities of
communication). All of the birds imported some current science (Cohen et al., 1993; Polis, 1995)
carbon within a range varying from 2% for ben- and the limitations of network analysis software
thos-eating birds to 93.5% for fish-eating birds. (e.g. ECOPATH II has a limit of 50 compartments).
The latter is largely the result of a flock of Amer- Gardner et al. (1982) and Cale (1995) addressed
Fig. 2. Effective trophic levels of compartments in the winter’s food web.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 109
aggregation strategies and their consequences. Al- partments can be ordered for evaluation and com-
though more needs to be learned, they concluded parison (Fig. 2). Effective trophic levels tended to
that aggregation errors may be minimized when cluster around integer values. Fifty per cent of the
aggregation involves (1) components with similar consumer compartments had effective trophic lev-
els at or near a level of 2 ( 5 2.32) signifying the
turnover times, (2) parallel components, (3) com-
ponents with common inputs and (4) components importance of herbivory and/or detritivory. There
with common outputs. Aggregation of compo- was then a small group which had values bridging
nents in series may be more problematic. This levels between 2 and 3 (three compartments with
would be especially true regarding trophic struc- levels from 2.41 to 2.74), a larger group near 3
ture. In the network described here, care was (ten from 2.90 to 3.22), a smaller group of four
taken to avoid aggregation of components in se- from 3.37 to 3.63, and four predators with levels
\ 3.88. Clustering near integer values may be the
ries (Luczkovich et al., 1997; submitted). The part
of the St. Marks food web wherein this may be of result of lack of trophic distinctions made for
greatest concern is the microbial community. smaller, prey organisms or a product of the aggre-
However, identifying the trophic structure of the gation of compartments. Small organisms tend to
microbial food web is a general problem of inves- be aggregated in food webs, whereas larger organ-
tigation (Pomeroy and Wiebe, 1988; Christian, isms are often identified to species or distinct
1994). Several microbial compartments were in- guilds (Cohen et al., 1993). This certainly was the
corporated into both the benthic and water case here. Although distinctions were made be-
column habitats at the St. Marks, and feeding tween bacteria, protozoans and meiofauna, these
among these compartments was included. The groups encompass considerable variability in diets
often hidden food web of microbes was thus made (Kemp, 1990; Sherr and Sherr, 1994). Fish and
explicit which expands the number of trophic birds, however, were largely grouped to include
levels; however an uncertainty about this part of one or a few species with similar diets and feeding
the web remains. habits.
The birds divide into a ‘herbivorous’ group (42)
3.2. Scaling by effecti6e trophic le6el. with level 2.28; benthic feeders (37) at 3.10; gulls
(40) at 3.41; and three compartments of fish eaters
Although others have computed effective (38), fish and crustacean eaters (39) and raptors
(41) ]3.88. This assumes that feeding off site is
trophic levels (e.g., Odum and Heald, 1975;
Ulanowicz, 1984; Johnson et al., 1995), effective comparable to that within the seagrass system.
trophic levels have not been used to structure This may be a reasonable assumption for birds
understanding of control and activity individual that feed in similar environments to the Halodule
compartments, as done here. The compartments community. Some, however, may feed differently.
listed in Table 1 were reordered and ranked by Gulls may feed in landfills and dump areas. Rap-
‘effective trophic level’ (Odum and Heald, 1975; tors may feed on prey from terrestrial environ-
Levine, 1980), as determined within ECOPATH II ments. The influence of such feeding is unknown.
and listed in Table 3 with selected output vari- Therefore, the conservative interpretation is that
ables. These variables include the effective trophic the effective trophic levels of organisms that im-
level; omnivory index; rates of consumption, pro- port considerable carbon are representative of
duction and respiration (mgC m − 2 d − 1), and their feeding within the system studied.
ecotrophic efficiency. The groups representing the Two studies of coastal Florida ecosystems pre-
highest trophic levels are listed at the top with sented and discussed effective trophic levels.
levels descending to primary producers and de- Odum and Heald (1975) evaluated the effective
tritus at the bottom. trophic structure of a mangrove ecosystem in
Effective trophic level represents the continu- south Florida. After correcting their values to
ous, rather than integer, trophic position of a make primary producers and detritus level 1,
compartment and is a good index by which com- many of the comparable taxa between our two
Table 3
110
Selected output values for winter 1994 in Goose Creek Bay, St. Marks National Wildlife Refuge, FL
Number Compartment name Effective Omnivory Productivity Consumption Respiration Ecotrophic
trophic level index (mgC m−2 d−1) efficiency
41 Raptors 4.32 1.69 2.77E−02 4.43E−01 3.05E−01 0.00
38 Fish-eating birds 4.00 0.53 5.54E−01 8.86E+00 6.09E+00 0.00
19 Gulf flounder & needlefish 3.91 0.00 2.14E−01 8.54E−01 4.64E−01 0.59
39 Fish and crustacean-eating birds 3.88 1.06 1.76E−02 2.81E−01 1.93E−01 0.00
20 Southern hake & sea robins 3.63 0.18 9.43E−02 3.75E−01 2.05E−01 0.10
40 Gulls 3.41 0.65 1.07E−01 1.72E+00 1.18E+00 0.00
28 Red drum 3.39 0.83 9.28E−02 3.71E−01 2.14E−01 0.09
21 Atlantic silverside & bay anchovies 3.37 0.25 8.30E−02 3.30E−01 1.79E−01 0.53
11 Blue crabs 3.22 0.14 9.98E−03 3.97E−02 1.00E−02 0.90
17 Catfish and stingrays 3.20 0.42 1.37E−01 5.49E−01 3.02E−01 0.62
37 Benthos-eating birds 3.10 0.03 2.84E−02 4.54E−01 3.12E−01 0.00
24 Gobies and blennies 3.10 0.01 3.41E−02 1.36E−01 7.50E−02 0.08
27 Pipefish and seahorses 3.10 0.04 3.76E−02 1.50E−01 8.30E−02 0.14
18 Tonguefish 3.05 0.01 2.16E−02 8.63E−02 4.80E−02 0.00
34 Predatory polychaetes 3.03 0.12 3.59E−01 1.43E+00 3.79E−01 0.97
16 Predatory shrimps 2.95 0.31 1.59E−01 6.38E−01 1.65E−01 1.17
26 Spot 2.91 0.30 2.84E+00 1.14E+01 6.27E+00 0.59
25 Pinfish 2.90 0.28 8.55E−02 3.42E−01 1.88E−01 0.25
2 Microfauna 2.74 0.26 1.88E+01 5.70E+01 2.68E+01 0.54
5 Microprotozoa 2.52 0.27 4.70E+00 1.47E+01 7.05E+00 0.11
23 Killifishes 2.41 0.48 2.84E−02 1.42E−01 8.50E−02 0.23
30 Predatory gastropods 2.32 0.23 2.81E+00 1.40E+01 4.25E+00 0.00
33 Deposit-feeding polychaetes 2.30 0.27 1.37E+00 9.14E+00 3.24E+00 0.92
42 Herbivorous ducks 2.28 0.21 5.25E−03 8.40E−02 5.80E−02 0.00
13 Brittle stars 2.27 0.26 9.59E−01 4.78E+00 1.45E+00 0.03
7 Suspension-feeding molluscs 2.23 0.22 4.93E−02 2.46E−01 7.40E−02 0.23
10 Omnivorous crabs 2.23 0.22 1.14E−01 5.71E−01 1.58E−01 1.00
3 Meiofauna 2.19 0.21 4.95E+01 1.65E+02 3.27E+01 0.31
14 Deposit-feeding peracaridan crustaceans 2.15 0.16 6.33E−01 4.20E+00 1.47E+00 0.63
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
36 Zooplankton 2.15 0.16 1.65E−01 8.25E−01 2.48E−01 0.81
8 Hermit crabs 2.12 0.12 5.96E−01 3.96E+00 1.43E+00 0.55
22 Sheepshead minnow 2.06 0.06 8.81E−02 5.88E−01 3.78E−01 1.02
29 Deposit-feeding gastropods 2.04 0.05 4.75E+00 3.17E+01 1.12E+01 1.08
35 Suspension-feeding polychaetes 2.01 0.01 8.72E−02 4.36E−01 1.31E−01 0.90
1 Benthic bacteria 2.00 0.00 6.56E+01 2.63E+02 1.71E+02 0.79
4 Bacterioplankton 2.00 0.00 1.66E+01 6.63E+01 4.98E+01 0.42
6 Epiphyte-grazing amphipods 2.00 0.00 7.08E−01 3.54E+00 1.07E+00 0.86
9 Spider crabs (herbivores) 2.00 0.00 1.77E−05 8.84E−05 0.00E+00 0.00
12 Isopods 2.00 0.00 4.03E−01 2.01E+00 6.06E−01 0.28
15 Herbivorous shrimps 2.00 0.00 8.10E−02 4.05E−01 1.25E−01 0.78
Table 3 (Continued)
Number Compartment name Effective Omnivory Productivity Consumption Respiration Ecotrophic
trophic level index (mgC m−2 d−1) efficiency
31 Epiphyte-grazing gastropods 2.00 0.00 1.05E−01 5.24E−01 1.58E−01 0.02
32 Other gastropods 2.00 0.00 1.70E−01 8.50E−01 2.56E−01 0.53
43 Halodule 1.00 0.00 9.93E+00 0.00E+00 0.18
44 Micro-epiphytes 1.00 0.00 1.95E+02 0.00E+00 0.02
45 Macro-epiphytes 1.00 0.00 1.62E+00 0.00E+00 0.34
46 Benthic algae 1.00 0.00 1.07E+02 0.00E+00 0.74
47 Phytoplankton 1.00 0.00 1.07E+02 0.00E+00 0.07
48 Detritus 1.00 0.42 0.82
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
111
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
112
Fig. 3. Productivity of compartments ordered according to effective trophic level in the winter’s food web.
systems had similar effective trophic levels. Bacte- compartments in two networks from marsh gut
ria and benthos clustered for both systems near ecosystems within the Crystal River, Florida, near
level 2. Their smaller, young fish tended to have St. Marks. His top carnivores had higher levels
values similar to those presented here, but these than reported here; over 40% of the compart-
fish groupings appeared to include more adult, ments having a level of 4.0 or greater. Few com-
small fish with higher values. This is reasonable as partments were at level 2. These results may be
Odum and Heald (1975) did not restrict them- hard to compare. Ulanowicz did not recycle mate-
selves to winter. Continuing up the food web, rial to detritus at level 1; as is done in ECOPATH II,
their top carnivores included raptors that had and as he has done in later analyses (Ulanowicz,
corrected effective trophic levels higher than for 1987). Thus, detritus in his networks had trophic
the St. Marks network. Thus, some of the differ- position greater than 2, which expanded the over-
ences may be the result of the timing and all range.
boundary conditions of the networks. Ulanowicz Others have calculated effective trophic level as
(1984) examined the effective trophic levels of 17 done here, including the authors represented in
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 113
the compilation of network analyses by Chris- and Sherr, 1994). Diets for detritivores were parti-
tensen and Pauly (1993). In general the range for tioned to include both detrital substratum and the
the St. Marks’ network was not unlike those associated microbial community. Fractions of diet
reported in those studies. A number of ecosystems among these compartments were largely in pro-
reported did not have levels above 4.0 (e.g. de la portion to relative biomass (Table 2). If detriti-
Cruz-Aguero, 1993; de Paula e Silva et al., 1993). vores fed only on detritus, they would have an
Whether the differences are the result of the food effective trophic level of 2.0. In this network,
webs or perceptions of them is not known. The detritivores have effective trophic levels greater
major differences are that the number of compart- than 2.0, reflecting these perceptions of the micro-
ments in the study presented here was larger than bial food web.
any reported in Christensen and Pauly (1993). Near steady state canonical or integer trophic
Also, the reports in the compilation did not in- levels provide a general trend of decreasing areal
clude explicit description of microbial processing productivity as trophic level increases (Lindeman,
of detritus. Hence the degree of aggregation for 1942; Ulanowicz and Kemp, 1979). Fig. 3 shows
many of the compartments reported here is less these general trends for effective trophic levels but
and the potential for more trophic steps may be with notable exceptions. Some groups with low
greater. Even when microbes have been explicitly effective trophic levels are rare and therefore have
included, effective trophic levels have rarely ex- low productivity values. Rare taxa at low trophic
ceeded 4.0 (Baird and Ulanowicz, 1989; Johnson levels, such as spider crabs (9), would not be out
et al., 1995). of the ordinary. Individual taxa (or compart-
Within ECOPATH II and NETWRK4, primary pro- ments) can be rare, but one expects that the
ducers and detritus are considered to have a composite biomass of a canonical trophic level
trophic level of 1, and therefore energy cycling and/or productivity would be greater than higher
can not be tracked through differential trophic levels. Compartments at high trophic levels would
positions of detritus; i.e. the trophic level detritus be expected to be found in lower abundance
is dependent on the level of the source organism and/or secondary productivity. Those at high lev-
(Burns, 1989). By assigning detritus different els with high productivity might be expected to be
trophic levels, different trophic structure, and quantitatively important controlling elements.
hence different effective trophic levels, are likely Their high rates of productivity would be associ-
to emerge (Burns et al., 1991). The theoretical ated with high rates of consumption and potential
value of unfolding energy cycling (Patten, 1985) is for top-down control. This feeding might impact
not argued. But the position of Baird and significantly on the community. Interestingly, this
Ulanowicz (1989) was adopted to accept what has potential for top-down control spread across taxa
become the more established calculation of de- from microscopic predators to overwintering wa-
tritus as level 1, as embodied within the available terfowl. Compartments with relatively high effec-
tive trophic levels ( ] 2.41) and relatively high
software.
rates of productivity (\ 0.2 mgC m − 2 d − 1) in-
In sampling for and constructing the current
network, an emphasis was placed on microbial cluded microprotozoans in the water column (5),
components associated with the detrital food web. microfauna in sediments (2), spot (26), predatory
If a consumer feeds on detritus, it also ingests the polychaetes (34), Gulf flounder and needlefish
associated microbial community. The microbial (19), and fish-eating birds (38).
community includes organisms that feed on the
3.3. Mixed trophic impact analysis
detrital substrate and on other members of the
community. Unfortunately, little is known about
the proportion of a detritivore’s diet that comes Mixed trophic impact analysis identifies the cu-
from the detrital substrate, the microbes feeding mulative impacts of each compartment on each
on the detrital substrate, and microbial predators other, whether positive or negative (Ulanowicz
(Lopez and Levinton, 1987; Kemp, 1990; Sherr and Puccia, 1990). Positive impacts promote ‘pop-
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
114
ulation’ growth and occur when one compartment analysis sums the impacts of each compartment
acts as a food source, reduces predation, or re- on each other across all trophic paths. To focus
duces competition on another compartment. Neg- on important interactions, only impacts greater
than or equal to 0.1 (i.e. a 10% effect) were
ative impacts, which reduce ‘population’ growth,
occur when one group acts as a competitor or a considered.
predator on another, or acts indirectly to promote Compartments were grouped along their effec-
competition or predation on another compart- tive trophic levels to determine the trends of posi-
ment. The impacts can be the result of direct tive (Fig. 4) and negative (Fig. 5) impacts. The
interaction between one compartment and an- number of positively impacted compartments was
other or the result of indirect interactions medi- generally inversely proportional to effective
ated through other compartment flows. Thus, the trophic level (Fig. 4). This may indicate bottom-
Fig. 4. Positive trophic impacts for the winter’s food web with compartments ordered by effective trophic level. The number of
impacted compartments represent those with coefficients \ or= 0.1 in the mixed trophic impact matrix.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 115
Fig. 5. Negative trophic impacts for the winter’s food web with compartments ordered by effective trophic level. The number of
impacted compartments represent those with coefficients B or= −0.1 in the mixed trophic impact matrix.
up control or the presence of multiple, diverse also had a positive impact on 4 compartments as
prey for the animal compartments, preventing a prey item and was the only ‘carnivore’ with so
strong linear foodchain linkages. All primary pro- many.
ducer compartments and detritus positively im- Compartments that provided the greatest num-
pacted multiple consumer compartments. Benthic bers of negative trophic impacts were among the
microalgae (46) and detritus (48) provided the herbivores and detritivores, but significant nega-
greatest potential for bottom-up control. Among tive impacts were caused by compartments with
largely herbivorous and detritivorous consumers, trophic levels above 3 (Fig. 5). Negative impacts
meiofauna (eight, at effective trophic level 2.19) by detritus and primary producers tended to be
and deposit-feeding gastropods (29, at effective through competition among primary producers or
trophic level 2.32), with their large biomasses, on the microbial community. Benthic bacteria (1)
positively impacted a disproportionate number of and meiofauna (3) provided the greatest numbers
compartments. However, epiphyte-grazing am- of negative impacts. These were through a num-
phipods (6), isopods (12), and deposit-feeding per- ber of mechanisms: competition with other con-
acaridan crustaceans (14) had relatively low sumers for detritus and benthic algal exudates
biomasses and positively impacted three or more (considered part of the detrital pool), consump-
groups. Spot (26, at effective trophic level 2.91) tion of benthic algal exudates and detritus de-
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
116
3.4. Feeding di6ersity
creasing their accumulation, and indirect effects
with other consumers. Spot (26) and gulls (40)
were the two compartments at higher trophic Another analysis provided by ECOPATH II is
levels that caused the most negative impacts. This that of an ‘omnivory index,’ the variance of the
was through their roles as predators and competi- effective trophic levels of a consumer’s preys
tors. Of the six groups with the highest effective (Christensen and Pauly, 1992). The diversity of
trophic levels, only one [fish and crustacean-eating trophic levels of prey fed upon by a predator
birds (39)] did not demonstrate negative impacts. increases with the index value. In Fig. 6 the
In fact three groups of birds (raptors, fish-eating omnivory indices for all compartments are listed
birds and gulls) demonstrated negative impacts in order of effective trophic level. There was a
within the community despite the fact that much trend for increased index values with increased
of their food had to be imported to achieve steady effective trophic level. Organisms at higher
state. trophic levels seemed to feed over a broader range
Fig. 6. Omnivory indexes of compartments ordered by effective trophic level.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 117
of levels than do lower levels. But this is not from the literature with modifications made for
without exceptions. At high trophic levels birds relative abundance of prey. Adjustments for
(37 – 42) generally have higher indices than fish steady state were based on diet distribution.
(17 – 28). Red drum (28) was an exception with the Effective trophic level was used as a metric for
third highest index. The high omnivory index of ordering compartments in the assessment of their
red drum may have resulted from the fact that attributes and interactions. As recognized by Lin-
juvenile and adult fish were pooled in the net- deman (1942), in a steady state system energy flow
work. Different fish life stages often have different decreases with increasing aggregate, canonical
diets, so the index reflects ontogenetic changes trophic level. Thus, as trophic level increases, the
(Livingston, 1980; Polis, 1995). At lower trophic energy flow of an average compartment at any
levels the indices were as low as 0. As in the effective trophic level decreases. Compartments
earlier discussion concerning the distribution of with attributes that diverge from this average
effective trophic levels, the low indices at low condition would be expected to have greater or
trophic levels was in part a result of the inability lesser influence on the food web. In the Halodule
to resolve diversity among microorganisms and community, consumer compartments comprise ef-
meiofauna. Such resolution would increase the fective trophic structure from 2.0 (herbivore/detri-
index, but this increase would in all probability tivore) to 4.32 (where 4.0 represents secondary
extend through higher levels that feed on these carnivory). The effective trophic levels of con-
groups. Thus the overall trend of increased om- sumers tend to aggregate near integer values, but
nivory indices with increased trophic level may the spread from integer values increases with in-
not be changed. creasing level. Based on productivity, several taxa
were found to be potentially important to energy
flow relative to their trophic position. These in-
4. Concluding remarks cluded protozoans in both the water column and
sediments, spot, predatory polychaetes, Gulf
Network analysis was conducted on a complex flounder and needlefish, and fish-eating birds. De-
and well articulated food web of a winter’s H. tritus and benthic microalgae were important
wrightii community in Goose Creek Bay, St. sources of food in the extended diets of many
Marks National Wildlife Refuge, FL. Unlike consumers. However, the importance of microal-
most such networks, much of the data used for gal production may have been underestimated
network construction came from sampling specific when dissolved photosynthate was modeled to
for that purpose. The strategy included field sam- pass through the detritus compartment, losing
pling the density and/or biomass of as many taxa track of the photosynthate’s origins within the
as possible, given time and personnel constraints. analyses. ‘Bottom-up’ control appeared important
Data from 4 samplings were averaged in this through mixed trophic impact analysis. The extent
process, two sites with replicate transects each of positive impacts decreased with increasing
sampled in January and February 1994. These trophic level. ‘Top-down’ control, as negative im-
data, diet estimates from fish stomach content pacts, appeared more limited to a few consumers
analyses, and selected process rates represented with inordinately large production relative to their
the core of the information base. This data base is trophic position. Ordering results from various
more specific in time and space than any other network analysis algorithms by effective trophic
used for foodweb network analysis. Furthermore, level proved useful in highlighting the potential
the complexity of the food web rivals or exceeds influence of different taxa to trophodynamics.
others in the literature (Baird and Ulanowicz, The energy flow through the winter’s Halodule
1989; Christensen, 1995; Ulanowicz et al., 1997). community is dominated by detritus and benthic
Most energetic processes were derived from litera- microalgae at the bottom and by waterfowl and
ture values using internally consistent rules. First piscivorous fish at the top. This pattern changes
approximations for other diet information came from winter to summer. SAV productivity in-
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
118
creases, and many of the birds emigrate (Zieman versity: Lori Beavers, Patrick Bishop, Giuseppe
and Zieman, 1989). Summer sampling demon- Castadelli, David Christian, Susan K. Dailey,
strated the immigration of other piscivorous fish Deborah Daniel, Matt Ferguson, Andrew
and sea turtles into the community (Luzckovich, Fletcher, Karen Halliday, Brian Harris, Jennifer
unpublished data). Furthermore, many of the fish Holmes, Martha Jones, Kris Lewis, Ed Moss,
captured in winter were juveniles. As they age, Chris Pullinger, and Garcy Ward. We also thank
many change diet ontogenetically (Livingston, William Rizzo and Hilary Neckles and colleagues
1980, 1984). The results of these changes in com- for their aid in data collection and analysis.
munity structure will undoubtedly change the Finally, we thank Dan Baird for his insights
trophic structure, a subject for further analysis. during his stay at ECU and Cristina Bondavalli,
Mike Erwin, Tommy Michot, Bill Rizzo, Bob
Ulanowicz, and Pace Wilber for reading and com-
Acknowledgements menting on a draft. This project was funded by
the US Fish and Wildlife Service through the
We thank the following people for help with National Wetlands Research Center, Lafayette,
field and laboratory work at East Carolina Uni- LA.
Appendix A. A list of the compartments and species
Compartment Compartment or common name Species or taxon pooled within a compartment
number
1 Benthic bacteria
2 Microfauna
3 Meiofauna
4 Bacterioplankton
5 Microprotozoa
6 Epiphyte grazing amphipods
Acunmindeutopus naglei
Ampithoe longimana
Caprella penantis
Cymadusa compta
Lembos rectangularis
Batea catharinensis
Elasmopus le6is
Melita sp.
Synchelidium sp.
Listriella barnardi
Lyssianopis alba
7 Suspension-feeding molluscs
Brachiodontes exustus
Chione cancellata
Argopecten irradians
Unident bi6al6es
Crepidula fornicata
Crepidula con6exa
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 119
8 Hermit crabs
Pagurus sp.
Pagurus mcglaughlini
9 Spider crabs Libinia dubia
10 Omnivorous crabs
Neopanope texana
Pinixia floridana
11 Blue crabs Callinectes sapidus
12 Isopods
Erichsionella sp.
Paracerces caudata
Edotea triloba
13 Brittle stars
Ophioderma bre6ispinum
14 Deposit feeding peracaridan crus-
traceans
Ampelisca sp.
Gammarus mucronatus
Cerapus tubularis
Corophium sp.
Detritivorous crustaceans
Unident. Cumacea
Unident. Tanaeid
Unident. ostracods
Mysidopsis
15 Herbivorous shrimp
Hippolyte zostericola
Alpheus normani
16 Predatory shrimp
Palaemonetes floridanus
Palaemonetes floridanus
Penaeus duoarum
Processa bermudiensis
17 Catfish and stingrays
Dasyatis sabina
Arius felis
18 Tonguefish Symphurus plagisua
19 Gulf flounder and needlefish
Paralichthyes albigutta
Strongylura marina
20 Southern hake and searobins
Urophycis floridana
Prionotus scitulus
Prionotus tribulus
21 Atlantic silversides and bay an-
chovy
Menidia beryllina
Anchoa mitchelli
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
120
22 Sheepshead minnow
23 Killifishes
Fundulus similis
Fundulus confluentus
Adinia xenica
24 Gobies and blennies
Microgobius gulosus
Gobiosoma robustum
25 Pinfish Lagodon rhomboides
26 Spot Leiostomus xanthurus
27 Pipefish and seahorses
Hippocampus zosterae
Syngnathus sco6elli
28 Red drum
(juveniles) Sciaenops ocellatus
(adults) Sciaenops ocellatus
29 Deposit-feeding gastropods
Acetocina candei
Swartziella catesbyana
Cadulus carolinesis
Haminoea succinea
Acteon punctostriatus
Oli6ella mutica
Truncatella pulchella
Nassarius 6ibex
30 Predatory gastropods
Unident. spirals
Urosalpinx perrugata
Unident. Nudibranchs
Opalia hotessieriana
Epitonium albidum
Terebra sp.
Polinices sp.
Busycon spiratum
Turbonilla dalli
Turbonilla hemphilli
Prunum (=Marginella) apicinum
Prunum (=Marginella) bellum
Prunum (=Marginella) aureocincta
Natica pusilla
Hylina 6eliei
Acanthocitona pygmaea
Odostomia seminuda
Seila adamsi
31 Epiphyte-grazing gastropods
Cerithium lutosum
Mitrella lunata
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 121
Solariella lamellosa
Anachis a6ara
32 Other gastropods
Mangelia plicosa
Hylina 6eliei
Jaspidella jaspidea
33 Deposit-feeding polychaetes
Aricidea sp.
Capitellidae
Cirratulidae
Maldanidae
Orbiniidae
Paraonidae
Pectanaridae
Syllidae
Amphitritidae
Spionidae
34 Predatory polychaetes and
nemertines
Glyceridae
Nereidae
Onuphidae
Hesionidae
Nemertines
35 Suspension-feeding polychaetes
Serpulidae
Sabellidae
36 Zooplankton
Acartia tonsa
Foraminifera
Harpacticoid
Nauplii1
Nauplii2
Nematode
Polychaete
Pycnogonid
37 Benthos-eating birds
Clapper Rail Rallus longirostris
Bufflehead Bucephala albeaola
Semi-palmated Plovers Charadrius semipalmatus
38 Fish-eating birds
Great Egret Casmerodius albus
Common Loon Ga6ia immer
Great Blue Heron Ardea herodias
Louisiana Heron Hydranassa tricolor
Red-Breasted Merganser Mergus serrator
Double-Crested Comorant Phalacrocorax carbo
Belted Kingfisher Megaceryle alcyon
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
122
39 Fish and crustacean eating birds
Hooded Merganser Lophodytes cucullatus
Willets Catoptrophorus semipalmatus
Greater Yellow Legs Tringa melanoleuca
40 Gulls and Terns
Forster’s Tern Sterna forsteri
Laughing Gull Larus atricilla
Herring Gull Larus argentatus
Ring-Billed Gull Larus delawarensis
41 Raptors
Bald Eagle Haliaeetus leucocephalus
Northern Harrier Circus cyaneus
42 Herbivorous ducks Anas discors
Blue-winged teal
43 Seagrass Halodule wrightii
44 Micro-epiphytes
45 Macro-epiphytes
46 Benthic algae
47 Phytoplankton
48 Detritus
References jacket, Oligoplites saurus and the spottail pinfish, Diplodus
holbrooki. Fishery Bull. US 70, 1111 – 1120.
Carr, W.E.S., Adams, C.A., 1973. Food habits of juvenile
Baird, D., Ulanowicz, R.E., 1989. The seasonal dynamics of
marine fishes occupying seagrass beds in the estuarine zone
the Chesapeake Bay ecosystem. Ecol. Monogr. 59, 329–
near Crystal River, Florida. Trans. Am. Fish. Soc. 102,
364.
511 – 540.
Baird, D., Ulanowicz, R.E., 1993. Comparative study on the
Christensen, V., 1995. Ecosystem maturity-towards quantifica-
trophic structure, cycling and ecosystem properties of four
tion. Ecol. Model. 77, 3 – 32.
tidal estuaries. Mar. Ecol. Prog. Ser. 99, 221–237.
Christensen, V., Pauly, D., 1992. A Guide to the ECOPATH II
Baird, D., Luczkovich, J.J., Christian, R.R., 1998. Assessment
Software System (Version 2.1). ICLARM, Manila,
of spatial and temporal variability in ecosystem attributes
Philippines.
of the St. Marks National Wildlife Refuge, Apalachee Bay,
Christensen, V., Pauly, D., 1993. Trophic Models of Aquatic
Florida. Estuar. Coast. Shelf Sci. 47, 329–349.
Ecosystems. ICLARM, Manila, Philippines.
Biujse, A.D., Van Eerden, M.R., Dekker, W., Van Densen,
Christian, R.R., 1994. Aggregation and disaggregation of mi-
W.L.T., 1993. Elements of a trophic model for Ijsselmer
crobial food webs. Microb. Ecol. 28, 327 – 330.
(The Netherlands), a shallow eutrophic lake. In: Chris-
Cohen, J.E., Beaver, R.A., Cousins, S.H., 1993. Improving
tensen, V., Pauly, D. (Eds.), Trophic models of aquatic
food webs. Ecology 74, 252 – 258.
ecosystems. ICLARM, Manila, Philippines, pp. 90–94.
de la Cruz-Aguero, G., 1993. A preliminary model of
Burns, T.P., 1989. Lindeman’s contradiction and the trophic
Mandinga lagoon, Veracruz, Mexico. In: Christensen, V.,
structure of ecosystems. Ecology 70, 1355–1362.
Pauly, D. (Eds.), Trophic Models of Aquatic Ecosystems.
Burns, T.P., Higashi, M., Wainright, S.C., Patten, B.C., 1991.
ICLARM, Manila, Philippines, pp. 193 – 196.
Trophic unfolding of a continental shelf energy–flow net-
de Paula e Silva, R., Sousa, M.I., Caramelo, A.M., 1993. The
work. Ecol. Model. 55, 1–26.
Maputo Bay ecosystem (Mozambique). In: Christensen,
Cale, W.G., 1995. Modeling aggregation: ecological perspec-
V., Pauly, D. Jr. (Eds.), Trophic Models of Aquatic
tives. In: B.C. Patten, S.E. Jørgensen (Eds.), Complex
Ecosystems. ICLARM, Manila, Philippines, pp. 214 –
Ecology: The Part-whole Relation in Ecosystems. Prentice-
223.
Hall PTR, Englewood Cliffs, NJ, USA, pp. 230–241.
Gardner, R.H., Cale, W.G., O’Neill, R.V., 1982. Robust anal-
Carr, W.E.S., Adams, C.A., 1972. Food habits of juvenile
marine fishes: evidence of cleaning habit in the leather- ysis of aggregation error. Ecology 63, 1771 – 1779.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 123
Mexico during winter. Report submitted to National
Heck, K.L. Jr., 1979. Some determinants of the composition
and abundance of motile macroinvertebrate species in Wetlands Research Center. US Fish and Wildlife Service,
tropical and temperate turtlegrass (Thalassia testudinum) Lafayette, LA.
meadows. J. Biogeog. 6, 183–200. Luczkovich, J.J., Rizzo, W.M., Neckles, H., Ward, G.,
Higgins, R.P., Thiel, H., 1988. Introduction to the Study of Christian, R.R., 1997. Including bacteria to birds in a
Meiofauna. Smithsonian Institution Press, Washington, food web: the sampling design for network analysis of a
DC. Halodule wrightii ecosystem in the Northeastern Gulf of
Johnson, C., Klumpp, D., Field, J., Bradbury, R., 1995. Mexico during winter. Ecol. Monogr. (submitted).
Carbon flux on coral reefs: effects of large shifts in com- Odum, W.E., Heald, E.J., 1975. The detritus-based food web
munity structure. Mar. Ecol. Prog. Ser. 126, 123–143. of an estuarine mangrove community. In: Cronin, L.E.
Jørgensen, S.E., Nielsen, S.N., Jørgensen, L., 1991. Hand- (Ed.), Estuarine Research, vol. 1. Academic Press, New
book of Ecological Parameters and Ecotoxicology, El- York, pp. 265 – 286.
sevier, Amsterdam. Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., Torres,
Kemp, P.F., 1990. The fate of benthic production. Rev. F. Jr., 1998. Fishing down marine food webs. Science
Aquatic Sci. 2, 109 – 124. 279, 860 – 863.
Landry, M.R., Hassett, R.P., 1982. Estimating the grazing Patten, B.C., 1985. Energy cycling in the ecosystem. Ecol.
impact of marine micro-zooplankton. Mar. Biol. 67, Model. 28, 1 – 71.
283 – 288. Peters, R.H., 1983. The Ecological Implications of Body
Leber, K.M., 1983. Feeding Ecology of Decapod Crus- Size. Cambridge University Press, New York.
taceans and the Influence of Vegetation on Foraging Polis, G.A., 1995. Complex food webs.. In: B.C. Patten, S.
Success in a Subtropical Seagrass Meadow. Ph.D. disser- E. Jørgensen (Eds.), Complex Ecology: The Part-whole
tation. The Florida State University, Tallahassee, FL. Relation in Ecosystems. Prentice Hall, Englewood Cliffs,
Levine, S., 1980. Several measures of trophic structure appli- NJ, USA, pp. 513 – 548.
cable to complex food webs. J. theor. Biol. 83, 195–207.
Pomeroy, L.R., Wiebe, W.J., 1988. Energetics of microbial
Lewis, F.G., 1984. Distribution of macrobenthic crustaceans
food webs. Hydrobiologia 159, 7 – 18.
associated with Thalassia, Halodule, and bare sand sub-
Sherr, E.B., Sherr, B.F., 1994. Bacterivory and herbivory:
strata. Mar. Ecol. Progr. Ser. 19, 101–113.
key roles of phagotrophic protists in pelagic food webs.
Lewis, F.G., Stoner, A.W., 1981. An examination of meth-
Microb. Ecol. 28, 223 – 235.
ods for sampling macrobenthos in seagrass meadowsm.
Stoner, A.W., 1979. Species-specific predation on amphipod
Bull. Mar. Sci. 31, 116–124.
crustacea by the pinfish Lagodon rhomboides: mediation
Lindeman, R.L., 1942. The trophic-dynamic aspect of ecol-
by macrophyte standing crop. Mar. Biol. 55, 201 – 207.
ogy. Ecology 23, 399 –418.
Stoner, A.W., 1980. The role of seagrass biomass in the
Livingston, R.J., 1980. Ontogenetic trophic relationships and
organization of benthic macrofaunal assemblages. Bull.
stress in a coastal seagrass system in Florida. In:
Mar. Sci. 30, 537 – 551.
Kennedy, V.S. (Ed.), Estuarine Perspectives. Academic
Tolan, J.M., Holt, S.A., Onuf, C.P., 1997. Distribution and
Press, New York, pp. 423–435.
community structure of ichthyoplankton in Laguna
Livingston, R.J., 1982. Trophic organization of fishes in a
Madre seagrass meadows: potential impact of seagrass
coastal seagrass system. Mar. Ecol. Progr. Ser. 7, 1–12.
species change. Estuaries 20, 450 – 464.
Livingston, R.J., 1984. Trophic response of fishes to habitat
Ulanowicz, R.E., 1984. Community measures of marine food
variability in coastal seagrass systems. Ecology 65, 1258–
networks and their possible applications. In: Fasham,
1275.
M.J.R. (Ed.), Flows of Energy and Materials in Marine
Lopez, G.R, Levinton, J.S., 1987. Ecology of deposit-feeding
Ecosystems: Theory and Practice. Plenum Press, New
animals in marine sediments. Quarterly Rev. Biol. 62,
York, pp. 23 – 47.
235 – 259.
Ulanowicz, R.E., 1987. NETWRK4: a Package of Computer
Luczkovich, J.J., 1987. The Patterns and Mechanisms of Se-
Algorithms to Analyze Ecological Flow Networks. Univ.
lective Feeding on Seagrass-Meadow Epifauna by Juve-
of Maryland, Chesapeake Biological Laboratory,
nile Pinfish, Lagodon rhomboides (Linnaeus). Ph.D.
Solomons.
dissertation. The Florida State University, Tallahassee,
Ulanowicz, R.E., Kemp, W.M., 1979. Toward canonical
FL.
trophic aggregations. Am. Nat. 114, 871 – 883.
Luczkovich, J.J., Stellwag, E.J., 1993. Isolation of cellulolytic
Ulanowicz, R.E., Puccia, C.J., 1990. Mixed trophic impacts
microbes from the intestinal tract of the pinfish, Lagodon
in ecosystems. Coenoses 5, 7 – 16.
rhomboides size-related changes in diet and microbial
Ulanowicz, R.E., Wulff, F., 1991. Comparing ecosystem
abundance. Mar. Biol. 116, 381–388.
structures: the Chesapeake Bay and the Baltic Sea. In:
Luczkovich, J.J., Rizzo, W.M., Neckles, H., Ward, G.,
Cole, J.J., Lovett, G.M., Findlay, S.E.G. (Eds.), Com-
Christian, R.R., 1997. Including bacteria to birds in a
parative Analyses of Ecosystems: Patterns, Mechanisms,
food web: the sampling design for network analysis of a
and Theories. Springer Verlag, New York, pp. 140 – 166.
Halodule wrightii ecosystem in the Northeastern Gulf of
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
124
Wulff, F., Field, J.G., Mann, K.H. (Eds.), 1989. Network
Ulanowicz, R.E., Bondavalli, C., Egnotovich, M.S., 1997.
Analysis in Marine Ecology: Methods and Applications,
Network Analysis of Trophic Dynamics in South Florida
Springer-Verlag. New York.
Ecosystems, FY 96: The Cypress Wetland Ecosystem. Ann.
Zieman, J.C., Zieman, R.T., 1989. The Ecology of the Sea-
Rept. to the USGS Biological Resources Division, U.
grass Meadows of the West Coast of Florida: a community
Miami, Coral Cables, FL.
profile, US Fish Widl. Serv. Biol. Rep. 85 (7.25).
.
Organizing and understanding a winter’s seagrass foodweb
network through effective trophic levels
Robert R. Christian *, Joseph J. Luczkovich
Biology Department, East Carolina Uni6ersity, Green6ille, NC 27858-4353, USA
Received 9 June 1998; accepted 5 January 1999
Abstract
Trophic structure of ecosystems is a unifying concept in ecology; however, the quantification of trophic level of
individual components has not received the attention one might expect. Ecosystem network analysis provides a
format to make several assessments of trophic structure of communities, including the effective trophic level (i.e.
non-integer) of these components. We applied network analysis to a Halodule wrightii community in Goose Creek
Bay, St. Marks National Wildlife Refuge, Florida, USA, during January and February 1994 where we sampled a wide
variety of taxa. Unlike most applications of network analysis, the field sampling design was specific for network
construction. From these data and literature values, we constructed and analyzed one of the most complex, highly
articulated and site specific foodweb networks to be done. Care was taken to structure the network to reflect best the
field data and ecology of populations within the requirements of analysis software. This involved establishing
internally consistent rules of data manipulation and compartment aggregation. Special attention was paid to the
microbial components of the food web. Consumer compartments comprised effective trophic levels from 2.0
(herbivore/detritivore) to 4.32 (where a level 4.0 represents ‘secondary carnivory’), and these values were used to
organize data interpretation. The effective trophic levels of consumers tended to aggregate near integer values, but the
spread from integer values increased with increasing level. Detritus and benthic microalgae acted as important sources
of food in the extended diets of many consumers. ‘Bottom-up’ control appeared important through mixed trophic
impact analysis, and the extent of positive impacts decreased with increasing trophic level. ‘Top-down’ control was
limited to a few consumers with relatively large production or biomass relative to their trophic position. Overall,
ordering results from various network analysis algorithms by effective trophic level proved useful in highlighting the
potential influence of different taxa to trophodynamics. Although the calculation of effective trophic level has been
available for some time, its application to the evaluation of other analyses has previously not received due
consideration. © 1999 Elsevier Science B.V. All rights reserved.
Keywords: Seagrass community; Network analysis; Effective trophic level; Carbon flow
* Corresponding author. Tel.: +1-252-3281835; fax: + 1-252-3284178.
E-mail address: christianr@mail.ecu.edu (R.R. Christian)
0304-3800/99/$ - see front matter © 1999 Elsevier Science B.V. All rights reserved.
PII: S 0 3 0 4 - 3 8 0 0 ( 9 9 ) 0 0 0 2 2 - 8
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
100
1. Introduction The seagrass communities along the coasts of
the southeastern United States and the Gulf of
Trophic structure is one of the primary ways by Mexico support substantial populations of ben-
which ecologists organize communities and thos, nekton and waterfowl (Zieman and Zieman,
ecosystems. Although trophic levels are often con- 1989). The primary producers in these communi-
sidered as discrete integers (Lindeman, 1942); in- ties may include several species of Submerged
dividual consumers, their populations or guilds Aquatic Vegetation (SAV), their epiphytes, phyto-
often feed across several trophic levels (Odum and plankton, benthic microalgae, and macroalgae.
Heald, 1975). Thus, populations or guilds may Birds and large fish represent important top con-
have ‘effective trophic levels’ that are fractional sumers, and the relative importance of each or-
(Odum and Heald, 1975; Levine, 1980). For ex- ganism may vary with season. In winter waterfowl
ample, a consumer that feeds as a herbivore (level are particularly abundant. The links between the
2) for 50% of its diet and as a primary carnivore primary producers and the top consumers are
(level 3) for 50% would have an effective trophic often poorly understood, with several trophic
level of 2.50. The calculations are done by both of steps between producer and top consumers. These
the most commonly used software packages for trophic steps may be mediated by microbes and
ecosystem network analysis; ECOPATH II (Chris- animals in both sediments and water column.
tensen and Pauly, 1992) and NETWRK4 (Ulanow- Ecosystem network analysis has been used to
icz, 1987). assess the foodweb interactions (Wulff et al.,
Although effective trophic level has been used 1989; Christensen and Pauly, 1993). The links
to characterize food webs, the full application of between primary producers and birds, a poten-
this classification has not been realized. Odum tially important consumer group, is rarely in-
and Heald (1975) used it to group various taxa cluded in complex network analyses (Baird and
into common feeding categories. Other re- Ulanowicz, 1993; Biujse et al., 1993). An effort
searchers have used it to compare trophic struc- was made here to include these consumers and
tures among ecosystems (e.g. Ulanowicz, 1984; evaluate their potential roles in trophodynamics.
Ulanowicz and Wulff, 1991). Recently, Pauly et The foundation for our research has been a
al. (1998) applied the concept to evaluate fishery priori collection of data to support the construc-
trends. It has even been used to examine theoreti- tion and analysis of a winter’s seagrass foodweb.
cal issues of energy flow (Burns, 1989). The em- Sampling was specifically designed for network
phasis in all of these studies has been at the construction and to be inclusive of the full range
ecosystem level. Little effort has been expended of trophic groupings (Luczkovich et al., 1997;
on the actual interaction of specific components submitted). We measured standing stocks of mi-
and their contribution to within-system regula- crobes, benthos, plankton, nekton, birds and or-
tion. Effective trophic level can be used as a ganic carbon as well as selected flows and diets.
scaling metric for other analyses to infer various The collection was a joint effort by us and staff of
attributes and contributions to trophodynamics. the National Wetlands Research Center, US Geo-
For example, populations with higher effective logical Survey (USGS). The focal ecosystem was
trophic levels would be expected to contribute less the seagrass communities within Goose Creek Bay
to the energetics of the ecosystem than those with of St. Marks National Wildlife Refuge, St. Marks,
lower levels. Deviations from this trend may indi- FL, USA. Further, Livingston and coworkers
cate that a consumer is particularly important or have amassed considerable information on the
unimportant to the food web. Also, the potential ecology of the northern Gulf of Mexico and its
for top-down or bottom-up control may be re- coastal and estuarine ecosystems (Heck, 1979;
lated to a population’s effective trophic level. In Stoner, 1979, 1980; Livingston, 1980, 1982, 1984;
this report effective trophic level was used to Lewis and Stoner, 1981; Leber, 1983; Lewis, 1984;
organize a seagrass food web and investigate these Luczkovich, 1987). From the field and laboratory
issues in that context. studies and the literature, foodweb networks were
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 101
constructed and analyzed with ECOPATH II for trophic position and potential importance of the
winter 1994. Three broad objectives were iden- different taxa in relation to position. Further-
tified with corresponding manuscripts. more, the issue of adapting field data for network
Luczkovich et al. (1997; submitted) addressed analysis is addressed.
sampling design related to the needs for network
construction following the guidelines of Cohen et
al. (1993). In Baird et al. (1998) networks were 2. Methods
reconstructed for analysis by NETWRK4 and evalu-
2.1. Sample site and design
ated for uncertainties of input and output vari-
ables with emphasis on systems-level attributes. In
the present paper the trophic structure of the Sampling was conducted from January and
system is evaluated with special attention to February 1994. Three sites were sampled in each
Fig. 1. Sample sites within St. Marks National Wildlife Refuge, St. Marks, Florida, USA. Numbered areas are sample sites.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
102
month: in January sites 1, 2, and 3; in February Zooplankton samples were obtained with a 90
mm mesh plankton net and preserved. In the
sites 1, 2, and 4 (Fig. 1). Sites 1 (Live Oak Island)
and 2 (Wakulla Beach) were within Goose Creek laboratory, the samples were sieved through a
Bay and similar in community structure and hy- series of screens, and the contents of each screen
drologic regime (Baird et al., 1998). Therefore, were counted as various taxa. Sieve fractions were
then dried at 60°C for 48 h and the biomass l − 1
only they were used for this network construction
and analysis. At each site sampling occurred at 3 in the original sample calculated, using the counts
transects, running perpendicular to the shore and to estimate proportional contribution of each tax-
extending through a Halodule wrightii community onomic group.
to approximately 150 m offshore. As a general Meiofauna were sampled by coring sediments,
minimum one sample was collected from each of preserved, and later separated from sediment,
the three transects within a site, but the different sorted, identified to the lowest possible taxon, and
variables measured required different sampling enumerated. Dimensions of representative organ-
procedures. Some variables were collected with isms were measured for conversion to biomass
much greater replication. (Higgins and Thiel, 1988).
To estimate the standing stock of macroinverte-
2.2. Sampling and sample analysis o6er6iew brates associated with seagrasses, 30 cores (7.62-
cm inside diameter) were taken per site (Lewis
Methods for field sampling and diet determina- and Stoner, 1981). The core samples were sieved
through 500 mm mesh in the field and placed into
tions are described in depth elsewhere by
Luczkovich et al. (1997, submitted) and Baird et jars with 10% formalin with rose bengal stain. In
al. (1998). Here we give an overview. the laboratory, animals in the sieved portions
Primary productivity and standing stocks of were sorted and identified to taxonomic groups.
primary producers were estimated largely by Polychaetes were identified to family level. Am-
USGS personnel directed by W. Rizzo and H. phipods, molluscs, decapod crustaceans and
Neckles. Ground cover along transects at each isopods were identified to species. Other inverte-
site was determined with periodic biomass sam- brate groups were identified as necessary. Biomass
pling. Biomass was divided by macrophyte species of each taxon was determined after drying. In the
(above- and below-ground), microepiphytes, and case of molluscs, ophuroids, polychaetes, isopods,
macrophytic algae. Benthic microalgal biomass decapods, and amphipods, ash-free dry masses
was estimated from chlorophyll a content in sur- were obtained by ashing representative samples
face layers of cores from each site. Phytoplankton and subtracting the mass of the remaining ash
biomass was estimated from aquatic chorophyll a from the dry mass. All ash-free dry masses were
concentrations. Benthic microalgal and phyto- converted to g carbon by multiplying by 0.45; for
plankton productivities were estimated from samples in which dry masses alone were deter-
changes in dissolved oxygen concentrations with mined, they were converted to g carbon by multi-
incubation in light and dark. plying by 0.40 (Jørgensen et al., 1991).
Benthic bacteria and sediment organic matter A technique developed for this study, the bar-
were sampled at each transect by coring to 5 cm, rier seine, and gill nets were used to sample fishes
and water samples were taken at each transect for and large mobile decapods at each station. All
dissolved and particulate carbon, bacterioplank- fishes caught in both gill nets and seines were
ton and planktonic microprotozoans. The densi- preserved in 10% formalin and taken back to the
ties of these organisms were estimated by lab where they were identified, counted, and
epiflourescence microscopy with appropriate weighed.
flourochromes. During each month at Wakulla Waterfowl standing stocks were estimated by
Beach, bacterioplankton growth and grazing rates surveys conducted during field campaigns and by
were estimated by modification of the method of D. Everette (The Florida State University, De-
Landry and Hassett (1982). partment of Biological Science). Everette made
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 103
five trips to Wakulla Beach and Live Oak Island were derived from allometric relationships to
from 4 February to 14 March 1994. On each body mass as summarized by Peters (1983) and
occasion and at each site, he counted birds within lowered to 75% of annual values to correct for
a 500×500 m2 area for 1 h. winter temperatures. QB values were then derived
To determine the structure of the diet matrix assuming set fractions of GE based on diet, where
required for ECOPATH II, dietary analyses were the fractions for detritivores, herbivores, omni-
conducted. Stomach content analysis was per- vores and carnivores were 0.15, 0.20, 0.20, and
formed on the most common fish species found in 0.25, respectively. Homeotherms (birds) were as-
the collections. Stomach contents of the fishes sumed to produce 1.5% of body mass per day
were analyzed following the sieve fractionation with food gross efficiencies of either 3 or 6% of
methodology of Carr and Adams (1972, 1973) as consumption. Lastly, diet distributions were esti-
modified by Luczkovich and Stellwag (1993). In mated for each consumer compartment from ei-
other cases, where fish samples were too small to ther gut analyses of field samples when available
conclude anything about diets, and for the inver- or the literature.
tebrate groups, estimates of dietary composition Ecosystem network analysis is actually a collec-
were obtained from the literature (see Baird et al., tion of mathematical algorithms to evaluate the
1998). structure of networks and ecosystems by infer-
ence. For this presentation interpretive efforts
2.3. Modelling and analysis approach concentrated on evaluating trophic structure and
the impacts of different organisms on trophody-
Biomass, given as mgC m − 2, was estimated for namics. ECOPATH II outputs of effective trophic
the various taxa collected in January and Febru- level (Levine, 1980), the mixed trophic impact
ary from sites 1 and 2. Estimation came from matrix (Ulanowicz and Puccia, 1990) and om-
either direct measurement of dry mass or conver- nivory index (Christensen and Pauly 1992) were
sion from density based on estimated dimensions used here. The algorithms for the three analyses
of the organisms. Taxa were then organized to are found in the cited references. Documentation
represent living compartments based on probable for these and other algorithms within the ECO-
diet and life history characteristics. The ‘detritus’ PATH II software are found in Christensen and
compartment was the sum of sediment organic Pauly (1992).
carbon and dissolved and estimated non-living
particulate carbon in the water column. The con-
version of volumetric to aerial data assumed a 3. Results and discussion
depth of 0.75 m for the water column and 5 cm
3.1. Input 6ariables and applying field data to
for the sediment.
ECOPATH II was used for network analysis network construction
(Christensen and Pauly, 1992). Version 2.1 was
used initially, but studies were completed with A listing of the compartments used in the food-
version 3.0 for Windows. The program required web network of a H. wrightii community in Goose
estimates of biomass (B) per compartment, Pro- Creek Bay averaged from January and February
ductivity: Biomass (PB), Consumption: Biomass 1994 is presented in Table 1. There are 48 com-
(QB), fraction of unassimilated food, and/or some partments. As with most representations of food
combined variable (e.g. Gross food conversion webs, the living compartments represent different
Efficiency (GE as PB/QB)). Some of these values degrees of aggregation (Cohen et al., 1993). Com-
were derived from field data, especially primary partments range from single species (e.g. 25 and
productivity of algae; but most came from litera- 26) to a few species (e.g. 24 and 38) to large
ture. Three important sources were Christensen groupings of taxa, especially of small organisms
and Pauly (1993), Jørgensen et al. (1991) and (e.g. 1 and 3). Similarities in diet and habitat are
Peters (1983). PB values for most poikilotherms the two main distinguishing characteristics for a
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
104
Table 1
networka
Summary of compartments for ECOPATH II
No. Compartment name Biomass PB (d) QB (d) Unassimilated Growth Fraction
(mgC/m2) food efficiency imported
1 Benthic bacteria 262.50 0.2500 1.0000 0.10 0.25
2 Microfauna 94.00 0.2000 0.6066 0.20 0.33
3 Meiofauna 1038.50 0.0476 0.1590 0.50 0.30
4 Bacterioplankton 10.90 1.5214 6.0855 0.00 0.25
5 Microprotozoa 4.70 1.0000 3.1250 0.20 0.32
6 Epiphyte-grazing amphipods 69.00 0.0103 0.0513 0.50 0.20
7 Suspension-feeding molluscs 6.76 0.0073 0.0364 0.50 0.19
8 Hermit crabs 178.52 0.0033 0.0222 0.50 0.14
9 Spider crabs (herbivores) 0.07 0.0002 0.0012 0.50 0.15
10 Omnivorous crabs 175.08 0.0007 0.0033 0.50 0.20
11 Blue crabs 12.74 0.0008 0.0031 0.50 0.25 0.1
12 Isopods 61.22 0.0066 0.0328 0.50 0.20
13 Brittle stars 370.83 0.0026 0.0129 0.50 0.20
14 Deposit-feeding peracaridan crustaceans 73.60 0.0086 0.0570 0.50 0.15
15 Herbivorous shrimps 24.58 0.0033 0.0165 0.50 0.20
16 Predatory shrimps 50.66 0.0031 0.0126 0.50 0.25
17 Catfish and stingrays 54.87 0.0025 0.0100 0.20 0.25 0.9
18 Tonguefish 1.44 0.0150 0.0599 0.20 0.25
19 Gulf flounder and needlefish 35.14 0.0061 0.0243 0.20 0.25
20 Southern hake and sea robins 9.34 0.0101 0.0402 0.20 0.25
21 Atlantic silverside and bay anchovies 7.90 0.0105 0.0418 0.20 0.26
22 Sheepshead minnow 8.39 0.0105 0.0700 0.20 0.16
23 Killifishes 2.26 0.0126 0.0628 0.20 0.21
24 Gobies and blennies 1.86 0.0183 0.0733 0.20 0.25
25 Pinfish 2.44 0.0351 0.1402 0.20 0.25
26 Spot 98.31 0.0289 0.1156 0.20 0.25
27 Pipefish and seahorses 1.41 0.0267 0.1066 0.20 0.25
28 Red drum 35.35 0.0026 0.0105 0.20 0.25 0.535
29 Deposit-feeding gastropods 974.93 0.0049 0.0325 0.50 0.15
30 Predatory gastropods 283.36 0.0099 0.0496 0.50 0.20
31 Epiphyte-grazing gastropods 6.46 0.0162 0.0811 0.50 0.20
32 Other gastropods 15.49 0.0110 0.0549 0.50 0.20
33 Deposit-feeding polychaetes 132.10 0.0104 0.0692 0.50 0.15
34 Predatory polychaetes 84.16 0.0043 0.0170 0.50 0.24
35 Suspension-feeding polychaetes 6.74 0.0129 0.0647 0.50 0.20
36 Zooplankton 2.50 0.0660 0.3301 0.50 0.20
37 Benthos-eating birds 1.89 0.0150 0.2400 0.25 0.06 0.02
38 Fish-eating birds 36.93 0.0150 0.2400 0.25 0.06 0.935
39 Fish and crustacean-eating birds 1.17 0.0150 0.2400 0.25 0.06 0.49
40 Gulls 7.17 0.0150 0.2400 0.25 0.06 0.855
41 Raptors 1.85 0.0150 0.2400 0.25 0.06 0.69
42 Herbivorous ducks 0.35 0.0150 0.2400 0.25 0.06 0.11
43 Halodule 4963.00 0.0020 0.0000 0.00
44 Micro-epiphytes 259.90 0.7500 0.0000 0.00
45 Macro-epiphytes 54.10 0.0300 0.0000 0.00
46 Benthic algae 1073.50 0.0997 0.0000 0.00
47 Phytoplankton 71.10 1.5000 0.0000 0.00
48 Detritus 369500
a
Average input values for winter 1994 in Goose Creek Bay, St. Marks National Wildlife Refuge, FL.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 105
compartmental grouping. Details of groupings are tropods. As calculated for input to the network,
described in Luczkovich et al. (1997) and summa- PB and QB values generally were inversely related
rized in Appendix A. Compartments are ordered to body size, being highest in the plankton.
numerically in the sequence with which data were Growth efficiencies for microbes and poikilo-
entered into ECOPATH II. In general, the order is therms ranged from 0.14 to 0.33. The lower effi-
microbes, benthic and epiphytic arthropods and ciency of 0.06 was used for birds.
bivalves, fish, gastropods and polychaetes, Although the populations of organisms in the
zooplankton, birds and at the end primary pro- field fluctuated over time, construction of a steady
ducers and detritus. state network was attempted. Steady state was
Most taxa were found during both months. considered achieved for any prey grouping that
Catfish and stingrays (17), tonguefish (18), gulf had an ecotrophic efficiency, i.e., fraction of pro-
flounder and needlefish (19), Atlantic silversides duction going to predation, harvest and export
and bay anchovies (21), and gobies and blennies (Christensen and Pauly, 1992) of 1 or less. To do
(24) were not found in January. Sheepshead min- this, one has several choices for modification of a
now (22), red drum (28), and killifish (23) were compartment’s attributes; biomass, parameter ra-
not found in February. Herbivorous spider crabs tios associated with metabolism, and food source.
(9) and herbivorous ducks (42) were not found in It was considered that the biomass data were the
January. All of these were included in the winter’s most reliable, as these were collected most di-
network. Values of zero were used for the month rectly. These data were not manipulated to
when the organisms were not present, and the achieve steady state. As described in the Methods
zeros were averaged with values obtained for the section, rules for parameter ratios were internally
month when the organisms were present. consistent for all, or at least related, groupings,
The 48 compartments are associated with 333 and these were not modified. Diet distributions,
individual transformations and transfers: 9 im- especially those from the literature, were subjected
ports, 47 respirations, 230 feeding pathways, and to the greatest manipulation because they were
47 returns to detritus (Tables 1 and 2). Cohen et considered to be a flexible parameter. Diets often
al. (1993) addressed the difficulty in presenting vary significantly across time and space in re-
large food webs through box and arrow diagrams, sponse to availability of different food items
indicating that graphical representations may be (Polis, 1995). If a prey grouping had an
too complicated to be meaningful. Therefore, in- ecotrophic efficiency greater than 1 (i.e. predation
formation used for network construction here is exceeded production in the network), the diet
given in tabular form as required for analysis by distributions of its predators were altered to re-
ECOPATH II. The input variables in Table 1 in- duce predation on it. After all reasonable alter-
clude biomass, PB, QB, their ratio as gross effi- ations of this kind, only three groups were
ciency, fraction of consumed food that is allowed to remain slightly overgrazed; predatory
unassimilated, and fraction of consumption im- shrimp (16), sheepshead minnow (22), and de-
ported from outside the site. The diet matrix in posit-feeding gastropods (29) (Table 3).
Table 2 includes feeding pathways from all food As a first assumption, most organisms were
sources to each consumer, as fractions of the considered to spend their time in the seagrass
consumer’s diet. All consumers contributed to community or in similar communities. Thus, there
detritus through mortality, egestion and excretion. was no import or export of material, unless dic-
As seen in Table 1 the largest biomass was in tated by the organism’s energetic balance and
detritus, primarily because of sediment organic biology. This appeared reasonable for many of
carbon. The primary producers, H. wrightii and the benthos and ichthyoplankton (Tolan et al.,
benthic microalgae had biomasses greater than 1997). During the process of adjusting ecotrophic
103 mgC m − 2. The only consumers to have efficiencies of prey, it became evident that some
biomasses around 103 mgC m − 2 were benthic predators could not be supported by the amount
fauna: i.e. meiofauna and deposit-feeding gas- of prey measured within the system. These preda-
106
Table 2
Diet matrix for winter 1994 in Goose Creek Bay, St. Marks National Wildlife Refuge, FL
Prey number Diet Composition as percentage of total ingestion of predator (designated by c)
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25
1 65 8 1 9.5 9.5
2 5 4 3 3
3 3 9 10 16.9 1 9 33 43
4 47 14
5 3 6
6 3 8 66 23.1 2 4 30 6
7 0.5 0.5
8 7 74 5
9
10 1 3 0.5 3.5
11
12 20 13
13 8.5
14 1.5 6.6 34 1 13 1 2 17 3
15 5 2.5 10 2.5
16 0.5 27 7
17
18
19
20 0.5
21
22
23
24
25 2 1
26 2 98 41 26 11
27 9
28
29 4.5 18
30
31
32
33 2.6 14.1 1.5 10 5.5
34 0.9 7 4 2
35 1.9 1 0.5
36 1 1 1 3 16
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
37
38
39
40
41
42
43 10 90 25 50 50 1 1
44 30 5 5 50 50 6
45 25 81 56
46 20 29 45 5 37.5 37.5 7 23
47 60 100 40 80 4.9
48 100 10 10 31 45 50 40 50 15 7 16 1
Table 2 (Continued)
Prey number Diet Composition as percentage of total ingestion of predator (designated by c )
26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42
1 1 0.5 1
2 1 1.5 1
3 60 40 1 18 42.5
4 7
5 5
6 2 27 7 1
7 0.5 0.5
8 5 15 3.5 6
9
10 5 1.5 3
11 0.5
12 12.5
13
14 1.5 12.5 7.5 0.5
15 0.5
16 0.5 0.5 0.5 0.5
17 19
18
19 1 8
20 1 2
21 1 2.5
22 1 0.5 0.8 2 2
23 1 1
24 1
25 1
26 8 5 40.2
27 0.5
28 2
29 30 38 70 23
30
31 0.5
32 0.5 1 1
33 8 2.2 7 5.5 6.5
34 2 0.8 3 3
35 0.5 0.5 0.5
36 0.5 2 1
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
37
38
39
40
41
42
43 62.5
44 100 100 3
45
46 25 2 48 25 99
47 9 88
48 49 69.5 50
107
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
108
tors were either nekton or birds, which are quite ican white pelicans that were frequently found at
able to leave the area within the time scale of the site for much of the winter. As a result of the
minutes to hours. They all had high biomasses importation of carbon associated with the steady
relative to their trophic position and high areal state assumption, the significance of bird feeding
consumption rates (Table 3). Blue crabs (11), on community structure could not be truly
catfish and rays (17), and red drum (28) were the quantified.
nekton groups, with the catfish and rays needing Aggregation of species into trophic guilds is
90% of their diet imported (Table 1). Rays may required for network analysis of most, if not all,
actually eat more frequently within the commu- natural ecosystems. This results from both the
nity, e.g. feeding on polychaetes not readily sam- fact that identification and characterization of all
pled by our techniques (P. Wilbur, personal species in an ecosystem are beyond the abilities of
communication). All of the birds imported some current science (Cohen et al., 1993; Polis, 1995)
carbon within a range varying from 2% for ben- and the limitations of network analysis software
thos-eating birds to 93.5% for fish-eating birds. (e.g. ECOPATH II has a limit of 50 compartments).
The latter is largely the result of a flock of Amer- Gardner et al. (1982) and Cale (1995) addressed
Fig. 2. Effective trophic levels of compartments in the winter’s food web.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 109
aggregation strategies and their consequences. Al- partments can be ordered for evaluation and com-
though more needs to be learned, they concluded parison (Fig. 2). Effective trophic levels tended to
that aggregation errors may be minimized when cluster around integer values. Fifty per cent of the
aggregation involves (1) components with similar consumer compartments had effective trophic lev-
els at or near a level of 2 ( 5 2.32) signifying the
turnover times, (2) parallel components, (3) com-
ponents with common inputs and (4) components importance of herbivory and/or detritivory. There
with common outputs. Aggregation of compo- was then a small group which had values bridging
nents in series may be more problematic. This levels between 2 and 3 (three compartments with
would be especially true regarding trophic struc- levels from 2.41 to 2.74), a larger group near 3
ture. In the network described here, care was (ten from 2.90 to 3.22), a smaller group of four
taken to avoid aggregation of components in se- from 3.37 to 3.63, and four predators with levels
\ 3.88. Clustering near integer values may be the
ries (Luczkovich et al., 1997; submitted). The part
of the St. Marks food web wherein this may be of result of lack of trophic distinctions made for
greatest concern is the microbial community. smaller, prey organisms or a product of the aggre-
However, identifying the trophic structure of the gation of compartments. Small organisms tend to
microbial food web is a general problem of inves- be aggregated in food webs, whereas larger organ-
tigation (Pomeroy and Wiebe, 1988; Christian, isms are often identified to species or distinct
1994). Several microbial compartments were in- guilds (Cohen et al., 1993). This certainly was the
corporated into both the benthic and water case here. Although distinctions were made be-
column habitats at the St. Marks, and feeding tween bacteria, protozoans and meiofauna, these
among these compartments was included. The groups encompass considerable variability in diets
often hidden food web of microbes was thus made (Kemp, 1990; Sherr and Sherr, 1994). Fish and
explicit which expands the number of trophic birds, however, were largely grouped to include
levels; however an uncertainty about this part of one or a few species with similar diets and feeding
the web remains. habits.
The birds divide into a ‘herbivorous’ group (42)
3.2. Scaling by effecti6e trophic le6el. with level 2.28; benthic feeders (37) at 3.10; gulls
(40) at 3.41; and three compartments of fish eaters
Although others have computed effective (38), fish and crustacean eaters (39) and raptors
(41) ]3.88. This assumes that feeding off site is
trophic levels (e.g., Odum and Heald, 1975;
Ulanowicz, 1984; Johnson et al., 1995), effective comparable to that within the seagrass system.
trophic levels have not been used to structure This may be a reasonable assumption for birds
understanding of control and activity individual that feed in similar environments to the Halodule
compartments, as done here. The compartments community. Some, however, may feed differently.
listed in Table 1 were reordered and ranked by Gulls may feed in landfills and dump areas. Rap-
‘effective trophic level’ (Odum and Heald, 1975; tors may feed on prey from terrestrial environ-
Levine, 1980), as determined within ECOPATH II ments. The influence of such feeding is unknown.
and listed in Table 3 with selected output vari- Therefore, the conservative interpretation is that
ables. These variables include the effective trophic the effective trophic levels of organisms that im-
level; omnivory index; rates of consumption, pro- port considerable carbon are representative of
duction and respiration (mgC m − 2 d − 1), and their feeding within the system studied.
ecotrophic efficiency. The groups representing the Two studies of coastal Florida ecosystems pre-
highest trophic levels are listed at the top with sented and discussed effective trophic levels.
levels descending to primary producers and de- Odum and Heald (1975) evaluated the effective
tritus at the bottom. trophic structure of a mangrove ecosystem in
Effective trophic level represents the continu- south Florida. After correcting their values to
ous, rather than integer, trophic position of a make primary producers and detritus level 1,
compartment and is a good index by which com- many of the comparable taxa between our two
Table 3
110
Selected output values for winter 1994 in Goose Creek Bay, St. Marks National Wildlife Refuge, FL
Number Compartment name Effective Omnivory Productivity Consumption Respiration Ecotrophic
trophic level index (mgC m−2 d−1) efficiency
41 Raptors 4.32 1.69 2.77E−02 4.43E−01 3.05E−01 0.00
38 Fish-eating birds 4.00 0.53 5.54E−01 8.86E+00 6.09E+00 0.00
19 Gulf flounder & needlefish 3.91 0.00 2.14E−01 8.54E−01 4.64E−01 0.59
39 Fish and crustacean-eating birds 3.88 1.06 1.76E−02 2.81E−01 1.93E−01 0.00
20 Southern hake & sea robins 3.63 0.18 9.43E−02 3.75E−01 2.05E−01 0.10
40 Gulls 3.41 0.65 1.07E−01 1.72E+00 1.18E+00 0.00
28 Red drum 3.39 0.83 9.28E−02 3.71E−01 2.14E−01 0.09
21 Atlantic silverside & bay anchovies 3.37 0.25 8.30E−02 3.30E−01 1.79E−01 0.53
11 Blue crabs 3.22 0.14 9.98E−03 3.97E−02 1.00E−02 0.90
17 Catfish and stingrays 3.20 0.42 1.37E−01 5.49E−01 3.02E−01 0.62
37 Benthos-eating birds 3.10 0.03 2.84E−02 4.54E−01 3.12E−01 0.00
24 Gobies and blennies 3.10 0.01 3.41E−02 1.36E−01 7.50E−02 0.08
27 Pipefish and seahorses 3.10 0.04 3.76E−02 1.50E−01 8.30E−02 0.14
18 Tonguefish 3.05 0.01 2.16E−02 8.63E−02 4.80E−02 0.00
34 Predatory polychaetes 3.03 0.12 3.59E−01 1.43E+00 3.79E−01 0.97
16 Predatory shrimps 2.95 0.31 1.59E−01 6.38E−01 1.65E−01 1.17
26 Spot 2.91 0.30 2.84E+00 1.14E+01 6.27E+00 0.59
25 Pinfish 2.90 0.28 8.55E−02 3.42E−01 1.88E−01 0.25
2 Microfauna 2.74 0.26 1.88E+01 5.70E+01 2.68E+01 0.54
5 Microprotozoa 2.52 0.27 4.70E+00 1.47E+01 7.05E+00 0.11
23 Killifishes 2.41 0.48 2.84E−02 1.42E−01 8.50E−02 0.23
30 Predatory gastropods 2.32 0.23 2.81E+00 1.40E+01 4.25E+00 0.00
33 Deposit-feeding polychaetes 2.30 0.27 1.37E+00 9.14E+00 3.24E+00 0.92
42 Herbivorous ducks 2.28 0.21 5.25E−03 8.40E−02 5.80E−02 0.00
13 Brittle stars 2.27 0.26 9.59E−01 4.78E+00 1.45E+00 0.03
7 Suspension-feeding molluscs 2.23 0.22 4.93E−02 2.46E−01 7.40E−02 0.23
10 Omnivorous crabs 2.23 0.22 1.14E−01 5.71E−01 1.58E−01 1.00
3 Meiofauna 2.19 0.21 4.95E+01 1.65E+02 3.27E+01 0.31
14 Deposit-feeding peracaridan crustaceans 2.15 0.16 6.33E−01 4.20E+00 1.47E+00 0.63
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
36 Zooplankton 2.15 0.16 1.65E−01 8.25E−01 2.48E−01 0.81
8 Hermit crabs 2.12 0.12 5.96E−01 3.96E+00 1.43E+00 0.55
22 Sheepshead minnow 2.06 0.06 8.81E−02 5.88E−01 3.78E−01 1.02
29 Deposit-feeding gastropods 2.04 0.05 4.75E+00 3.17E+01 1.12E+01 1.08
35 Suspension-feeding polychaetes 2.01 0.01 8.72E−02 4.36E−01 1.31E−01 0.90
1 Benthic bacteria 2.00 0.00 6.56E+01 2.63E+02 1.71E+02 0.79
4 Bacterioplankton 2.00 0.00 1.66E+01 6.63E+01 4.98E+01 0.42
6 Epiphyte-grazing amphipods 2.00 0.00 7.08E−01 3.54E+00 1.07E+00 0.86
9 Spider crabs (herbivores) 2.00 0.00 1.77E−05 8.84E−05 0.00E+00 0.00
12 Isopods 2.00 0.00 4.03E−01 2.01E+00 6.06E−01 0.28
15 Herbivorous shrimps 2.00 0.00 8.10E−02 4.05E−01 1.25E−01 0.78
Table 3 (Continued)
Number Compartment name Effective Omnivory Productivity Consumption Respiration Ecotrophic
trophic level index (mgC m−2 d−1) efficiency
31 Epiphyte-grazing gastropods 2.00 0.00 1.05E−01 5.24E−01 1.58E−01 0.02
32 Other gastropods 2.00 0.00 1.70E−01 8.50E−01 2.56E−01 0.53
43 Halodule 1.00 0.00 9.93E+00 0.00E+00 0.18
44 Micro-epiphytes 1.00 0.00 1.95E+02 0.00E+00 0.02
45 Macro-epiphytes 1.00 0.00 1.62E+00 0.00E+00 0.34
46 Benthic algae 1.00 0.00 1.07E+02 0.00E+00 0.74
47 Phytoplankton 1.00 0.00 1.07E+02 0.00E+00 0.07
48 Detritus 1.00 0.42 0.82
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
111
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
112
Fig. 3. Productivity of compartments ordered according to effective trophic level in the winter’s food web.
systems had similar effective trophic levels. Bacte- compartments in two networks from marsh gut
ria and benthos clustered for both systems near ecosystems within the Crystal River, Florida, near
level 2. Their smaller, young fish tended to have St. Marks. His top carnivores had higher levels
values similar to those presented here, but these than reported here; over 40% of the compart-
fish groupings appeared to include more adult, ments having a level of 4.0 or greater. Few com-
small fish with higher values. This is reasonable as partments were at level 2. These results may be
Odum and Heald (1975) did not restrict them- hard to compare. Ulanowicz did not recycle mate-
selves to winter. Continuing up the food web, rial to detritus at level 1; as is done in ECOPATH II,
their top carnivores included raptors that had and as he has done in later analyses (Ulanowicz,
corrected effective trophic levels higher than for 1987). Thus, detritus in his networks had trophic
the St. Marks network. Thus, some of the differ- position greater than 2, which expanded the over-
ences may be the result of the timing and all range.
boundary conditions of the networks. Ulanowicz Others have calculated effective trophic level as
(1984) examined the effective trophic levels of 17 done here, including the authors represented in
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 113
the compilation of network analyses by Chris- and Sherr, 1994). Diets for detritivores were parti-
tensen and Pauly (1993). In general the range for tioned to include both detrital substratum and the
the St. Marks’ network was not unlike those associated microbial community. Fractions of diet
reported in those studies. A number of ecosystems among these compartments were largely in pro-
reported did not have levels above 4.0 (e.g. de la portion to relative biomass (Table 2). If detriti-
Cruz-Aguero, 1993; de Paula e Silva et al., 1993). vores fed only on detritus, they would have an
Whether the differences are the result of the food effective trophic level of 2.0. In this network,
webs or perceptions of them is not known. The detritivores have effective trophic levels greater
major differences are that the number of compart- than 2.0, reflecting these perceptions of the micro-
ments in the study presented here was larger than bial food web.
any reported in Christensen and Pauly (1993). Near steady state canonical or integer trophic
Also, the reports in the compilation did not in- levels provide a general trend of decreasing areal
clude explicit description of microbial processing productivity as trophic level increases (Lindeman,
of detritus. Hence the degree of aggregation for 1942; Ulanowicz and Kemp, 1979). Fig. 3 shows
many of the compartments reported here is less these general trends for effective trophic levels but
and the potential for more trophic steps may be with notable exceptions. Some groups with low
greater. Even when microbes have been explicitly effective trophic levels are rare and therefore have
included, effective trophic levels have rarely ex- low productivity values. Rare taxa at low trophic
ceeded 4.0 (Baird and Ulanowicz, 1989; Johnson levels, such as spider crabs (9), would not be out
et al., 1995). of the ordinary. Individual taxa (or compart-
Within ECOPATH II and NETWRK4, primary pro- ments) can be rare, but one expects that the
ducers and detritus are considered to have a composite biomass of a canonical trophic level
trophic level of 1, and therefore energy cycling and/or productivity would be greater than higher
can not be tracked through differential trophic levels. Compartments at high trophic levels would
positions of detritus; i.e. the trophic level detritus be expected to be found in lower abundance
is dependent on the level of the source organism and/or secondary productivity. Those at high lev-
(Burns, 1989). By assigning detritus different els with high productivity might be expected to be
trophic levels, different trophic structure, and quantitatively important controlling elements.
hence different effective trophic levels, are likely Their high rates of productivity would be associ-
to emerge (Burns et al., 1991). The theoretical ated with high rates of consumption and potential
value of unfolding energy cycling (Patten, 1985) is for top-down control. This feeding might impact
not argued. But the position of Baird and significantly on the community. Interestingly, this
Ulanowicz (1989) was adopted to accept what has potential for top-down control spread across taxa
become the more established calculation of de- from microscopic predators to overwintering wa-
tritus as level 1, as embodied within the available terfowl. Compartments with relatively high effec-
tive trophic levels ( ] 2.41) and relatively high
software.
rates of productivity (\ 0.2 mgC m − 2 d − 1) in-
In sampling for and constructing the current
network, an emphasis was placed on microbial cluded microprotozoans in the water column (5),
components associated with the detrital food web. microfauna in sediments (2), spot (26), predatory
If a consumer feeds on detritus, it also ingests the polychaetes (34), Gulf flounder and needlefish
associated microbial community. The microbial (19), and fish-eating birds (38).
community includes organisms that feed on the
3.3. Mixed trophic impact analysis
detrital substrate and on other members of the
community. Unfortunately, little is known about
the proportion of a detritivore’s diet that comes Mixed trophic impact analysis identifies the cu-
from the detrital substrate, the microbes feeding mulative impacts of each compartment on each
on the detrital substrate, and microbial predators other, whether positive or negative (Ulanowicz
(Lopez and Levinton, 1987; Kemp, 1990; Sherr and Puccia, 1990). Positive impacts promote ‘pop-
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
114
ulation’ growth and occur when one compartment analysis sums the impacts of each compartment
acts as a food source, reduces predation, or re- on each other across all trophic paths. To focus
duces competition on another compartment. Neg- on important interactions, only impacts greater
than or equal to 0.1 (i.e. a 10% effect) were
ative impacts, which reduce ‘population’ growth,
occur when one group acts as a competitor or a considered.
predator on another, or acts indirectly to promote Compartments were grouped along their effec-
competition or predation on another compart- tive trophic levels to determine the trends of posi-
ment. The impacts can be the result of direct tive (Fig. 4) and negative (Fig. 5) impacts. The
interaction between one compartment and an- number of positively impacted compartments was
other or the result of indirect interactions medi- generally inversely proportional to effective
ated through other compartment flows. Thus, the trophic level (Fig. 4). This may indicate bottom-
Fig. 4. Positive trophic impacts for the winter’s food web with compartments ordered by effective trophic level. The number of
impacted compartments represent those with coefficients \ or= 0.1 in the mixed trophic impact matrix.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 115
Fig. 5. Negative trophic impacts for the winter’s food web with compartments ordered by effective trophic level. The number of
impacted compartments represent those with coefficients B or= −0.1 in the mixed trophic impact matrix.
up control or the presence of multiple, diverse also had a positive impact on 4 compartments as
prey for the animal compartments, preventing a prey item and was the only ‘carnivore’ with so
strong linear foodchain linkages. All primary pro- many.
ducer compartments and detritus positively im- Compartments that provided the greatest num-
pacted multiple consumer compartments. Benthic bers of negative trophic impacts were among the
microalgae (46) and detritus (48) provided the herbivores and detritivores, but significant nega-
greatest potential for bottom-up control. Among tive impacts were caused by compartments with
largely herbivorous and detritivorous consumers, trophic levels above 3 (Fig. 5). Negative impacts
meiofauna (eight, at effective trophic level 2.19) by detritus and primary producers tended to be
and deposit-feeding gastropods (29, at effective through competition among primary producers or
trophic level 2.32), with their large biomasses, on the microbial community. Benthic bacteria (1)
positively impacted a disproportionate number of and meiofauna (3) provided the greatest numbers
compartments. However, epiphyte-grazing am- of negative impacts. These were through a num-
phipods (6), isopods (12), and deposit-feeding per- ber of mechanisms: competition with other con-
acaridan crustaceans (14) had relatively low sumers for detritus and benthic algal exudates
biomasses and positively impacted three or more (considered part of the detrital pool), consump-
groups. Spot (26, at effective trophic level 2.91) tion of benthic algal exudates and detritus de-
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
116
3.4. Feeding di6ersity
creasing their accumulation, and indirect effects
with other consumers. Spot (26) and gulls (40)
were the two compartments at higher trophic Another analysis provided by ECOPATH II is
levels that caused the most negative impacts. This that of an ‘omnivory index,’ the variance of the
was through their roles as predators and competi- effective trophic levels of a consumer’s preys
tors. Of the six groups with the highest effective (Christensen and Pauly, 1992). The diversity of
trophic levels, only one [fish and crustacean-eating trophic levels of prey fed upon by a predator
birds (39)] did not demonstrate negative impacts. increases with the index value. In Fig. 6 the
In fact three groups of birds (raptors, fish-eating omnivory indices for all compartments are listed
birds and gulls) demonstrated negative impacts in order of effective trophic level. There was a
within the community despite the fact that much trend for increased index values with increased
of their food had to be imported to achieve steady effective trophic level. Organisms at higher
state. trophic levels seemed to feed over a broader range
Fig. 6. Omnivory indexes of compartments ordered by effective trophic level.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 117
of levels than do lower levels. But this is not from the literature with modifications made for
without exceptions. At high trophic levels birds relative abundance of prey. Adjustments for
(37 – 42) generally have higher indices than fish steady state were based on diet distribution.
(17 – 28). Red drum (28) was an exception with the Effective trophic level was used as a metric for
third highest index. The high omnivory index of ordering compartments in the assessment of their
red drum may have resulted from the fact that attributes and interactions. As recognized by Lin-
juvenile and adult fish were pooled in the net- deman (1942), in a steady state system energy flow
work. Different fish life stages often have different decreases with increasing aggregate, canonical
diets, so the index reflects ontogenetic changes trophic level. Thus, as trophic level increases, the
(Livingston, 1980; Polis, 1995). At lower trophic energy flow of an average compartment at any
levels the indices were as low as 0. As in the effective trophic level decreases. Compartments
earlier discussion concerning the distribution of with attributes that diverge from this average
effective trophic levels, the low indices at low condition would be expected to have greater or
trophic levels was in part a result of the inability lesser influence on the food web. In the Halodule
to resolve diversity among microorganisms and community, consumer compartments comprise ef-
meiofauna. Such resolution would increase the fective trophic structure from 2.0 (herbivore/detri-
index, but this increase would in all probability tivore) to 4.32 (where 4.0 represents secondary
extend through higher levels that feed on these carnivory). The effective trophic levels of con-
groups. Thus the overall trend of increased om- sumers tend to aggregate near integer values, but
nivory indices with increased trophic level may the spread from integer values increases with in-
not be changed. creasing level. Based on productivity, several taxa
were found to be potentially important to energy
flow relative to their trophic position. These in-
4. Concluding remarks cluded protozoans in both the water column and
sediments, spot, predatory polychaetes, Gulf
Network analysis was conducted on a complex flounder and needlefish, and fish-eating birds. De-
and well articulated food web of a winter’s H. tritus and benthic microalgae were important
wrightii community in Goose Creek Bay, St. sources of food in the extended diets of many
Marks National Wildlife Refuge, FL. Unlike consumers. However, the importance of microal-
most such networks, much of the data used for gal production may have been underestimated
network construction came from sampling specific when dissolved photosynthate was modeled to
for that purpose. The strategy included field sam- pass through the detritus compartment, losing
pling the density and/or biomass of as many taxa track of the photosynthate’s origins within the
as possible, given time and personnel constraints. analyses. ‘Bottom-up’ control appeared important
Data from 4 samplings were averaged in this through mixed trophic impact analysis. The extent
process, two sites with replicate transects each of positive impacts decreased with increasing
sampled in January and February 1994. These trophic level. ‘Top-down’ control, as negative im-
data, diet estimates from fish stomach content pacts, appeared more limited to a few consumers
analyses, and selected process rates represented with inordinately large production relative to their
the core of the information base. This data base is trophic position. Ordering results from various
more specific in time and space than any other network analysis algorithms by effective trophic
used for foodweb network analysis. Furthermore, level proved useful in highlighting the potential
the complexity of the food web rivals or exceeds influence of different taxa to trophodynamics.
others in the literature (Baird and Ulanowicz, The energy flow through the winter’s Halodule
1989; Christensen, 1995; Ulanowicz et al., 1997). community is dominated by detritus and benthic
Most energetic processes were derived from litera- microalgae at the bottom and by waterfowl and
ture values using internally consistent rules. First piscivorous fish at the top. This pattern changes
approximations for other diet information came from winter to summer. SAV productivity in-
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
118
creases, and many of the birds emigrate (Zieman versity: Lori Beavers, Patrick Bishop, Giuseppe
and Zieman, 1989). Summer sampling demon- Castadelli, David Christian, Susan K. Dailey,
strated the immigration of other piscivorous fish Deborah Daniel, Matt Ferguson, Andrew
and sea turtles into the community (Luzckovich, Fletcher, Karen Halliday, Brian Harris, Jennifer
unpublished data). Furthermore, many of the fish Holmes, Martha Jones, Kris Lewis, Ed Moss,
captured in winter were juveniles. As they age, Chris Pullinger, and Garcy Ward. We also thank
many change diet ontogenetically (Livingston, William Rizzo and Hilary Neckles and colleagues
1980, 1984). The results of these changes in com- for their aid in data collection and analysis.
munity structure will undoubtedly change the Finally, we thank Dan Baird for his insights
trophic structure, a subject for further analysis. during his stay at ECU and Cristina Bondavalli,
Mike Erwin, Tommy Michot, Bill Rizzo, Bob
Ulanowicz, and Pace Wilber for reading and com-
Acknowledgements menting on a draft. This project was funded by
the US Fish and Wildlife Service through the
We thank the following people for help with National Wetlands Research Center, Lafayette,
field and laboratory work at East Carolina Uni- LA.
Appendix A. A list of the compartments and species
Compartment Compartment or common name Species or taxon pooled within a compartment
number
1 Benthic bacteria
2 Microfauna
3 Meiofauna
4 Bacterioplankton
5 Microprotozoa
6 Epiphyte grazing amphipods
Acunmindeutopus naglei
Ampithoe longimana
Caprella penantis
Cymadusa compta
Lembos rectangularis
Batea catharinensis
Elasmopus le6is
Melita sp.
Synchelidium sp.
Listriella barnardi
Lyssianopis alba
7 Suspension-feeding molluscs
Brachiodontes exustus
Chione cancellata
Argopecten irradians
Unident bi6al6es
Crepidula fornicata
Crepidula con6exa
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 119
8 Hermit crabs
Pagurus sp.
Pagurus mcglaughlini
9 Spider crabs Libinia dubia
10 Omnivorous crabs
Neopanope texana
Pinixia floridana
11 Blue crabs Callinectes sapidus
12 Isopods
Erichsionella sp.
Paracerces caudata
Edotea triloba
13 Brittle stars
Ophioderma bre6ispinum
14 Deposit feeding peracaridan crus-
traceans
Ampelisca sp.
Gammarus mucronatus
Cerapus tubularis
Corophium sp.
Detritivorous crustaceans
Unident. Cumacea
Unident. Tanaeid
Unident. ostracods
Mysidopsis
15 Herbivorous shrimp
Hippolyte zostericola
Alpheus normani
16 Predatory shrimp
Palaemonetes floridanus
Palaemonetes floridanus
Penaeus duoarum
Processa bermudiensis
17 Catfish and stingrays
Dasyatis sabina
Arius felis
18 Tonguefish Symphurus plagisua
19 Gulf flounder and needlefish
Paralichthyes albigutta
Strongylura marina
20 Southern hake and searobins
Urophycis floridana
Prionotus scitulus
Prionotus tribulus
21 Atlantic silversides and bay an-
chovy
Menidia beryllina
Anchoa mitchelli
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
120
22 Sheepshead minnow
23 Killifishes
Fundulus similis
Fundulus confluentus
Adinia xenica
24 Gobies and blennies
Microgobius gulosus
Gobiosoma robustum
25 Pinfish Lagodon rhomboides
26 Spot Leiostomus xanthurus
27 Pipefish and seahorses
Hippocampus zosterae
Syngnathus sco6elli
28 Red drum
(juveniles) Sciaenops ocellatus
(adults) Sciaenops ocellatus
29 Deposit-feeding gastropods
Acetocina candei
Swartziella catesbyana
Cadulus carolinesis
Haminoea succinea
Acteon punctostriatus
Oli6ella mutica
Truncatella pulchella
Nassarius 6ibex
30 Predatory gastropods
Unident. spirals
Urosalpinx perrugata
Unident. Nudibranchs
Opalia hotessieriana
Epitonium albidum
Terebra sp.
Polinices sp.
Busycon spiratum
Turbonilla dalli
Turbonilla hemphilli
Prunum (=Marginella) apicinum
Prunum (=Marginella) bellum
Prunum (=Marginella) aureocincta
Natica pusilla
Hylina 6eliei
Acanthocitona pygmaea
Odostomia seminuda
Seila adamsi
31 Epiphyte-grazing gastropods
Cerithium lutosum
Mitrella lunata
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 121
Solariella lamellosa
Anachis a6ara
32 Other gastropods
Mangelia plicosa
Hylina 6eliei
Jaspidella jaspidea
33 Deposit-feeding polychaetes
Aricidea sp.
Capitellidae
Cirratulidae
Maldanidae
Orbiniidae
Paraonidae
Pectanaridae
Syllidae
Amphitritidae
Spionidae
34 Predatory polychaetes and
nemertines
Glyceridae
Nereidae
Onuphidae
Hesionidae
Nemertines
35 Suspension-feeding polychaetes
Serpulidae
Sabellidae
36 Zooplankton
Acartia tonsa
Foraminifera
Harpacticoid
Nauplii1
Nauplii2
Nematode
Polychaete
Pycnogonid
37 Benthos-eating birds
Clapper Rail Rallus longirostris
Bufflehead Bucephala albeaola
Semi-palmated Plovers Charadrius semipalmatus
38 Fish-eating birds
Great Egret Casmerodius albus
Common Loon Ga6ia immer
Great Blue Heron Ardea herodias
Louisiana Heron Hydranassa tricolor
Red-Breasted Merganser Mergus serrator
Double-Crested Comorant Phalacrocorax carbo
Belted Kingfisher Megaceryle alcyon
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
122
39 Fish and crustacean eating birds
Hooded Merganser Lophodytes cucullatus
Willets Catoptrophorus semipalmatus
Greater Yellow Legs Tringa melanoleuca
40 Gulls and Terns
Forster’s Tern Sterna forsteri
Laughing Gull Larus atricilla
Herring Gull Larus argentatus
Ring-Billed Gull Larus delawarensis
41 Raptors
Bald Eagle Haliaeetus leucocephalus
Northern Harrier Circus cyaneus
42 Herbivorous ducks Anas discors
Blue-winged teal
43 Seagrass Halodule wrightii
44 Micro-epiphytes
45 Macro-epiphytes
46 Benthic algae
47 Phytoplankton
48 Detritus
References jacket, Oligoplites saurus and the spottail pinfish, Diplodus
holbrooki. Fishery Bull. US 70, 1111 – 1120.
Carr, W.E.S., Adams, C.A., 1973. Food habits of juvenile
Baird, D., Ulanowicz, R.E., 1989. The seasonal dynamics of
marine fishes occupying seagrass beds in the estuarine zone
the Chesapeake Bay ecosystem. Ecol. Monogr. 59, 329–
near Crystal River, Florida. Trans. Am. Fish. Soc. 102,
364.
511 – 540.
Baird, D., Ulanowicz, R.E., 1993. Comparative study on the
Christensen, V., 1995. Ecosystem maturity-towards quantifica-
trophic structure, cycling and ecosystem properties of four
tion. Ecol. Model. 77, 3 – 32.
tidal estuaries. Mar. Ecol. Prog. Ser. 99, 221–237.
Christensen, V., Pauly, D., 1992. A Guide to the ECOPATH II
Baird, D., Luczkovich, J.J., Christian, R.R., 1998. Assessment
Software System (Version 2.1). ICLARM, Manila,
of spatial and temporal variability in ecosystem attributes
Philippines.
of the St. Marks National Wildlife Refuge, Apalachee Bay,
Christensen, V., Pauly, D., 1993. Trophic Models of Aquatic
Florida. Estuar. Coast. Shelf Sci. 47, 329–349.
Ecosystems. ICLARM, Manila, Philippines.
Biujse, A.D., Van Eerden, M.R., Dekker, W., Van Densen,
Christian, R.R., 1994. Aggregation and disaggregation of mi-
W.L.T., 1993. Elements of a trophic model for Ijsselmer
crobial food webs. Microb. Ecol. 28, 327 – 330.
(The Netherlands), a shallow eutrophic lake. In: Chris-
Cohen, J.E., Beaver, R.A., Cousins, S.H., 1993. Improving
tensen, V., Pauly, D. (Eds.), Trophic models of aquatic
food webs. Ecology 74, 252 – 258.
ecosystems. ICLARM, Manila, Philippines, pp. 90–94.
de la Cruz-Aguero, G., 1993. A preliminary model of
Burns, T.P., 1989. Lindeman’s contradiction and the trophic
Mandinga lagoon, Veracruz, Mexico. In: Christensen, V.,
structure of ecosystems. Ecology 70, 1355–1362.
Pauly, D. (Eds.), Trophic Models of Aquatic Ecosystems.
Burns, T.P., Higashi, M., Wainright, S.C., Patten, B.C., 1991.
ICLARM, Manila, Philippines, pp. 193 – 196.
Trophic unfolding of a continental shelf energy–flow net-
de Paula e Silva, R., Sousa, M.I., Caramelo, A.M., 1993. The
work. Ecol. Model. 55, 1–26.
Maputo Bay ecosystem (Mozambique). In: Christensen,
Cale, W.G., 1995. Modeling aggregation: ecological perspec-
V., Pauly, D. Jr. (Eds.), Trophic Models of Aquatic
tives. In: B.C. Patten, S.E. Jørgensen (Eds.), Complex
Ecosystems. ICLARM, Manila, Philippines, pp. 214 –
Ecology: The Part-whole Relation in Ecosystems. Prentice-
223.
Hall PTR, Englewood Cliffs, NJ, USA, pp. 230–241.
Gardner, R.H., Cale, W.G., O’Neill, R.V., 1982. Robust anal-
Carr, W.E.S., Adams, C.A., 1972. Food habits of juvenile
marine fishes: evidence of cleaning habit in the leather- ysis of aggregation error. Ecology 63, 1771 – 1779.
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124 123
Mexico during winter. Report submitted to National
Heck, K.L. Jr., 1979. Some determinants of the composition
and abundance of motile macroinvertebrate species in Wetlands Research Center. US Fish and Wildlife Service,
tropical and temperate turtlegrass (Thalassia testudinum) Lafayette, LA.
meadows. J. Biogeog. 6, 183–200. Luczkovich, J.J., Rizzo, W.M., Neckles, H., Ward, G.,
Higgins, R.P., Thiel, H., 1988. Introduction to the Study of Christian, R.R., 1997. Including bacteria to birds in a
Meiofauna. Smithsonian Institution Press, Washington, food web: the sampling design for network analysis of a
DC. Halodule wrightii ecosystem in the Northeastern Gulf of
Johnson, C., Klumpp, D., Field, J., Bradbury, R., 1995. Mexico during winter. Ecol. Monogr. (submitted).
Carbon flux on coral reefs: effects of large shifts in com- Odum, W.E., Heald, E.J., 1975. The detritus-based food web
munity structure. Mar. Ecol. Prog. Ser. 126, 123–143. of an estuarine mangrove community. In: Cronin, L.E.
Jørgensen, S.E., Nielsen, S.N., Jørgensen, L., 1991. Hand- (Ed.), Estuarine Research, vol. 1. Academic Press, New
book of Ecological Parameters and Ecotoxicology, El- York, pp. 265 – 286.
sevier, Amsterdam. Pauly, D., Christensen, V., Dalsgaard, J., Froese, R., Torres,
Kemp, P.F., 1990. The fate of benthic production. Rev. F. Jr., 1998. Fishing down marine food webs. Science
Aquatic Sci. 2, 109 – 124. 279, 860 – 863.
Landry, M.R., Hassett, R.P., 1982. Estimating the grazing Patten, B.C., 1985. Energy cycling in the ecosystem. Ecol.
impact of marine micro-zooplankton. Mar. Biol. 67, Model. 28, 1 – 71.
283 – 288. Peters, R.H., 1983. The Ecological Implications of Body
Leber, K.M., 1983. Feeding Ecology of Decapod Crus- Size. Cambridge University Press, New York.
taceans and the Influence of Vegetation on Foraging Polis, G.A., 1995. Complex food webs.. In: B.C. Patten, S.
Success in a Subtropical Seagrass Meadow. Ph.D. disser- E. Jørgensen (Eds.), Complex Ecology: The Part-whole
tation. The Florida State University, Tallahassee, FL. Relation in Ecosystems. Prentice Hall, Englewood Cliffs,
Levine, S., 1980. Several measures of trophic structure appli- NJ, USA, pp. 513 – 548.
cable to complex food webs. J. theor. Biol. 83, 195–207.
Pomeroy, L.R., Wiebe, W.J., 1988. Energetics of microbial
Lewis, F.G., 1984. Distribution of macrobenthic crustaceans
food webs. Hydrobiologia 159, 7 – 18.
associated with Thalassia, Halodule, and bare sand sub-
Sherr, E.B., Sherr, B.F., 1994. Bacterivory and herbivory:
strata. Mar. Ecol. Progr. Ser. 19, 101–113.
key roles of phagotrophic protists in pelagic food webs.
Lewis, F.G., Stoner, A.W., 1981. An examination of meth-
Microb. Ecol. 28, 223 – 235.
ods for sampling macrobenthos in seagrass meadowsm.
Stoner, A.W., 1979. Species-specific predation on amphipod
Bull. Mar. Sci. 31, 116–124.
crustacea by the pinfish Lagodon rhomboides: mediation
Lindeman, R.L., 1942. The trophic-dynamic aspect of ecol-
by macrophyte standing crop. Mar. Biol. 55, 201 – 207.
ogy. Ecology 23, 399 –418.
Stoner, A.W., 1980. The role of seagrass biomass in the
Livingston, R.J., 1980. Ontogenetic trophic relationships and
organization of benthic macrofaunal assemblages. Bull.
stress in a coastal seagrass system in Florida. In:
Mar. Sci. 30, 537 – 551.
Kennedy, V.S. (Ed.), Estuarine Perspectives. Academic
Tolan, J.M., Holt, S.A., Onuf, C.P., 1997. Distribution and
Press, New York, pp. 423–435.
community structure of ichthyoplankton in Laguna
Livingston, R.J., 1982. Trophic organization of fishes in a
Madre seagrass meadows: potential impact of seagrass
coastal seagrass system. Mar. Ecol. Progr. Ser. 7, 1–12.
species change. Estuaries 20, 450 – 464.
Livingston, R.J., 1984. Trophic response of fishes to habitat
Ulanowicz, R.E., 1984. Community measures of marine food
variability in coastal seagrass systems. Ecology 65, 1258–
networks and their possible applications. In: Fasham,
1275.
M.J.R. (Ed.), Flows of Energy and Materials in Marine
Lopez, G.R, Levinton, J.S., 1987. Ecology of deposit-feeding
Ecosystems: Theory and Practice. Plenum Press, New
animals in marine sediments. Quarterly Rev. Biol. 62,
York, pp. 23 – 47.
235 – 259.
Ulanowicz, R.E., 1987. NETWRK4: a Package of Computer
Luczkovich, J.J., 1987. The Patterns and Mechanisms of Se-
Algorithms to Analyze Ecological Flow Networks. Univ.
lective Feeding on Seagrass-Meadow Epifauna by Juve-
of Maryland, Chesapeake Biological Laboratory,
nile Pinfish, Lagodon rhomboides (Linnaeus). Ph.D.
Solomons.
dissertation. The Florida State University, Tallahassee,
Ulanowicz, R.E., Kemp, W.M., 1979. Toward canonical
FL.
trophic aggregations. Am. Nat. 114, 871 – 883.
Luczkovich, J.J., Stellwag, E.J., 1993. Isolation of cellulolytic
Ulanowicz, R.E., Puccia, C.J., 1990. Mixed trophic impacts
microbes from the intestinal tract of the pinfish, Lagodon
in ecosystems. Coenoses 5, 7 – 16.
rhomboides size-related changes in diet and microbial
Ulanowicz, R.E., Wulff, F., 1991. Comparing ecosystem
abundance. Mar. Biol. 116, 381–388.
structures: the Chesapeake Bay and the Baltic Sea. In:
Luczkovich, J.J., Rizzo, W.M., Neckles, H., Ward, G.,
Cole, J.J., Lovett, G.M., Findlay, S.E.G. (Eds.), Com-
Christian, R.R., 1997. Including bacteria to birds in a
parative Analyses of Ecosystems: Patterns, Mechanisms,
food web: the sampling design for network analysis of a
and Theories. Springer Verlag, New York, pp. 140 – 166.
Halodule wrightii ecosystem in the Northeastern Gulf of
R.R. Christian, J.J. Luczko6ich / Ecological Modelling 117 (1999) 99–124
124
Wulff, F., Field, J.G., Mann, K.H. (Eds.), 1989. Network
Ulanowicz, R.E., Bondavalli, C., Egnotovich, M.S., 1997.
Analysis in Marine Ecology: Methods and Applications,
Network Analysis of Trophic Dynamics in South Florida
Springer-Verlag. New York.
Ecosystems, FY 96: The Cypress Wetland Ecosystem. Ann.
Zieman, J.C., Zieman, R.T., 1989. The Ecology of the Sea-
Rept. to the USGS Biological Resources Division, U.
grass Meadows of the West Coast of Florida: a community
Miami, Coral Cables, FL.
profile, US Fish Widl. Serv. Biol. Rep. 85 (7.25).
.