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).




                              .