Effects of nutrient enrichment and small predator density on seagrass ecosystems: An experimental assessment

1041

Limnol. Oceanogr., 45(5), 2000, 1041–1057q 2000, by the American Society of Limnology and Oceanography, Inc.

Effects of nutrient enrichment and small predator density on seagrass ecosystems: Anexperimental assessment

K. L. Heck, Jr.1

Dauphin Island Sea Lab, Department of Marine Science, University of South Alabama, 101 Bienville Boulevard, DauphinIsland, Alabama 36528

J. R. PennockDauphin Island Sea Lab, Department of Biology, University of Alabama, 101 Bienville Boulevard, Dauphin Island,Alabama 36528

J. F. ValentineDauphin Island Sea Lab, Department of Marine Science, University of South Alabama, 101 Bienville Boulevard, DauphinIsland, Alabama 36528

L. D. CoenSouth Carolina Department of Natural Resources, Marine Resources Research Institute, P.O. Box 12559, Charleston, SouthCarolina 29422

S. A. SklenarDauphin Island Sea Lab, Department of Marine Science, University of South Alabama, 101 Bienville Boulevard, DauphinIsland, Alabama 36528

Abstract

We used a field experiment to assess the individual and combined effects of removing top predators and enrichingwater column nutrients (nitrogen-N and phosphorus-P) on seagrass ecosystem structure and function. Experimentswere conducted in turtlegrass (Thalassia testudinum) habitats in St. Joseph Bay, FL, an aquatic preserve in thenorthern Gulf of Mexico that exhibits low ambient nutrient concentrations and contains abundant populations ofsmall crustacean and gastropod mesograzers. We stocked 7.0 m2 enclosures with elevated (;4–83 ambient) den-sities of juvenile pinfish (Lagodon rhomboides), the dominant fish species in local seagrass habitats, to simulate thefirst-order effects of large predator reductions, and we used an in situ delivery system to supplement N and P to;33 ambient levels in nutrient addition treatments. Monthly determinations of water column nutrients and Chl a,along with measurements of the biomass and abundance of leaf epiphytes and seagrass production, biomass, andshoot and leaf densities were used to evaluate the relative effects of manipulating nutrient supply and altering foodweb structure.

In contrast to our expectations, results showed few significant nutrient effects, or fish 3 nutrient enrichmenteffects on any of the parameters measured. However, there were many significant fish effects, most of which wereunexpected. As predicted, increased pinfish density reduced mesograzer numbers significantly. Not anticipated,however, was the reduced epiphyte biomass in fish enclosure treatments, apparently brought about by the pinfishconsuming significant amounts of epiphytes as well as mesograzers. This reduction in epiphyte biomass producedpositive indirect effects on seagrass biomass, shoot number, and rates of primary productivity in pinfish enclosuretreatments.

Our results also showed important top-down effects in determining the composition and abundance of seagrass-associated plants and animals in this pristine environment. Although we did not observe simple trophic cascades,most likely because pinfish fed at more than one trophic level, and because the dense seagrass prevented smallgrazers from being reduced to low numbers, pinfish produced important changes in the epibiota as well as theseagrasses themselves. These data, while contrasting with studies reporting significant negative nutrient enrichmenteffects on seagrasses, support the results of recent experimental studies in showing that: (1) small grazers can oftencontrol the abundance of epiphytes; and (2) it is unlikely that a full understanding of the consequences of nutrientenrichment for seagrass ecosystems can be gained without knowing how grazer population are regulated.

1 Corresponding author ([email protected]).

AcknowledgmentsWe thank Leah Gregory, Katherine Canter, Brad Peterson, Paul Bologna, Patric Harper, David Webb, Kirsten Walker, along with a host

of other graduate students, for help in field sampling and cage maintenance, Carolyn Wood for manuscript preparation, and Thad Murdochfor drafting Fig. 1. Support was provided by the National Science Foundation Alabama EPSCoR Program, the Dauphin Island Sea Laband the University of South Alabama. DISL Contribution 317.

1042 Heck et al.

Nutrient enrichment and harvesting of large, predatoryfishes are two of the most common anthropogenic pertur-bations in coastal ecosystems. Each has been shown to pro-duce dramatic changes in ecosystem structure and function.For example, increased nutrient loading of estuaries maylead to noxious algal blooms, increased sedimentation of or-ganic material, and ultimately oxygen depletion in bottomwaters (see Officer et al. 1984; Nixon 1995 for an overview).Nutrient enrichment can produce other undesirable effects inaquatic systems, including the loss of submerged aquaticvegetation (SAV) as a result of shading by rapidly prolifer-ating algal epibionts, whose growth is hypothesized to out-strip the ability of grazers to control them (van Montfranset al. 1982; 1984; Orth and van Montfrans 1984; Bronmark1985; Twilley et al. 1985; Howard and Short 1986; Kaiser1989; Tomasko and Lapointe 1991). After SAV disappears,phytoplankton often come to dominate primary productionat very high levels of nutrient loading (see review by Duarte1995). Such ‘‘bottom-up’’ control of ecosystem structure andfunction by nutrient supply suggests that in eutrophic sys-tems, the potential for a shift from a macrophyte-based to aplankton-based food web is great (Orth and Moore 1983;Cambridge and McComb 1984; Giesen et al. 1990).

It is also known that small invertebrate grazers (i.e., me-sograzers) play an important role in controlling epiphyticalgal abundance (Howard 1982; van Montfrans et al. 1982;Hootsmans and Vermaat 1985). Recent experimental inves-tigations of the influence of mesograzers (e.g., amphipods,isopods, and small gastropods) on epiphyte abundance innutrient enriched conditions have found that in most instanc-es (excluding those with the most extreme examples of nu-trient enrichment) grazers can control epiphyte abundanceand prevent seagrass decline (Neckles et al. 1993; Williamsand Ruckelshaus 1993; Short et al. 1995). Why this grazercontrol has not been frequently reported in nature is unclear,but it suggests that factors influencing grazer abundancemight also be involved in determining the degree to whichnutrient enrichment impacts SAV habitats. It also suggeststhat a better understanding of the factors that control me-sograzer abundance is needed before we can fully understandthe role of nutrient supply in determining the health of sea-grass resources.

Reductions in abundance and changes in the compositionof predatory fish guilds can also produce profound changesin aquatic systems. Examples include changes in prey habitatutilization and large shifts in prey composition due to re-ductions in the abundance of predators (Brooks and Dodson1965; Paine 1966; Dayton 1971; Zaret 1980; Coen et al.1981; Mittelbach 1984; Werner and Gilliam 1984; Main1985; 1987). In the pelagic zone of temperate lakes, removalof large predators can trigger a trophic cascade that leads togreater abundances of small fish species, shifts from large tosmall zooplankton species, and shifts from palatable to graz-er-resistant phytoplankton species (Shapiro and Wright 1984;see also reviews by Carpenter et al. 1985; 1987). In littoralfreshwater habitats, experimentally reducing predator densityindirectly led to dramatic changes in SAV abundance (Mar-tin et al. 1992; Lodge et al. 1994). Similarly, in marine kelpforests overharvesting of predatory sea otters led to largeincreases in the density of their sea urchin prey, which sub-

sequently brought about the loss of kelps as they were over-grazed by the urchins (Estes and Palmisano 1974; Duggins1980). On coral reefs, the removal of herbivorous fishes hasled to the proliferation of algae with concomitant loss ofcoral cover as algal species begin to monopolize availablespace (Hughes 1994).

It is now well documented that the ocean’s predators havebeen greatly reduced by fishing, and many popular articleshave increased the public’s awareness and concern about theconsequences of removing great numbers of predators fromthe world’s oceans (Parfit 1995; Safina 1995). As the Na-tional Academy of Sciences (1995) reported, the drastic re-ductions in many species of preferred fishes may be exten-sive enough to endanger the function of entire marineecosystems. This report ranked fishing activities as the mostserious threat the oceans now face.

Based on these observations, we hypothesized that the in-creasing harvest of fish predators in coastal waters (e.g.,large, warm temperate sciaenid species, such as red drum[Sciaenops ocellata], and spotted sea trout [Cynoscion ne-bulosus]) could produce ‘‘top-down’’ effects, in some re-spects similar to those observed in the littoral zone in lakes(Martin et al. 1992; Lodge et al. 1994), that could ultimatelyshift seagrass dominated ecosystems to less productive un-vegetated bottoms. Such a shift could be mediated by thefollowing sequence of changes after removing most largefish predators: (1) increases in the density of small predatoryfishes (e.g., pinfish [Lagodon rhomboides], pigfish [Ortho-pristis chrysoptera], and silver perch [Bairdiella chrysura])as they are released from predation; (2) decreases in theabundance of the mesograzer prey of these small fishes (e.g.,amphipods, gastropods, and caridean shrimp); (3) increasesin the epibiont abundance on the seagrasses as their meso-grazer consumers decline in number; and (4) eventual dis-appearance of the seagrasses as they become overgrown byepibionts. It is important to note that the results predictedby manipulating top predators are similar to those of excessnutrient additions: namely, the loss of seagrass habitat as aresult of epiphytic overgrowth. In addition, the effect of re-ducing large predator populations could increase the rate ofseagrass loss in moderately enriched habitats much morerapidly than nutrient addition alone.

Until relatively recently, few investigators have studiedboth ‘‘top-down’’ and/or ‘‘bottom-up’’ responses of thewhole SAV food web (e.g., Carpenter and Lodge 1986; Mar-tin et al. 1992; Lodge et al. 1994; Bronmark and Weisner1996), and we are unaware of any who have attempted suchwork in coastal waters. Important differences are to be ex-pected between marine and previously well-studied fresh-water systems (Heck and Crowder 1991). For example, thecumulative effects of shifts in macrophyte or large predatorabundance are predicted to be more profound in small,‘‘closed’’ systems such as ponds and less important in larger,‘‘open’’ systems such as rivers (Power 1992) and estuaries(Heck and Crowder 1991). It is also possible that species-rich marine communities, with many omnivorous taxa, maybe less susceptible to ‘‘top-down’’ effects than less diversefresh water communities (Strong 1992; Polis and Strong1996).

Here we report on an in situ simulation of the separate

1043Nutrients and predator manipulations

Fig. 1. Study site location in St. Joseph Bay, Florida.

and interactive effects of altering the abundance of large fishpredators and nutrient concentrations in seagrass-dominatedcoastal marine ecosystems. Our objectives were to developa mechanistic understanding of the indirect effects that mayresult from these most common perturbations of coastal eco-systems, and ultimately, to predict when these systems couldbe expected to shift between macrophyte- and phytoplank-ton-dominated states.

Study site

St. Joseph Bay, Florida, located in the northeastern Gulfof Mexico (29.88N, 85.38W), is a soft-bottom polyhaline es-tuarine system (Fig. 1) with no significant source of fresh-water input. Seagrass meadows are dominated by turtlegrass,Thalassia testudinum, but also contain shoalgrass, Halodulewrightii, and manatee grass, Syringodium filiforme, alongwith unvegetated sand habitats, in the shallow (,2 m) areas.Salinities generally range from 30 to 36 PSU annually(Stewart and Gorsline 1962; pers. observation) but extremevalues range from 26 to 43 PSU. Temperatures vary annuallyfrom approximately 8–308C (pers. observation) and the

mean tidal range is 0.5 m (Rudloe 1985). Our previous un-published measurements have shown low water column nu-trient levels ranging from 0.01–2.73 mM for nitrate, 0.3–2.6mM for ammonium, 0–15.35 mM for silicate and 0–0.14 mMfor phosphate (unpubl. data). Overall, Chl a (Chl a) valuesare low, and range from 0.17–6.16 mg L21 (unpubl. data).St. Joseph Bay is a semienclosed lagoon type system char-acterized by low energy current regimes. During the summerof 1994, current velocities measured at our study site rangedfrom 0 to 7.5 cm s21, (mean 5 2.8 cm s21) , as estimatedby an InterOcean System S4 current meter.

Methods

Experimental design—Twenty-four 7 m2 round enclosureswere erected parallel to shore at depths of approximately 1m in an area of dense turtlegrass to test the effects of nutrientenrichment and small predator density on community struc-ture and function at several trophic levels. Each enclosurewas made of plastic net (1.3 3 1.8 cm mesh) held in placeby a metal reinforcing rod frame. Bird netting tops (1.9 cmmesh) maximized light passage and prevented large fish

1044 Heck et al.

Fig. 2. Dissolution rates vs. time for nitrate 1 nitrite, ammo-nium and phosphate at varying temperatures for (2 inch diameter)PVC tubes containing 500 g premeasured OsmocoteTM.

Table 1. Nutrient release rates (mmol tube21 d21) for PVC tubes containing Osmocotey both in the field and in the laboratory. Fieldrelease rates are calculated from weight loss measurements while laboratory measurements were carried out at 158C and 258C for comparisonpurposes.

DatesTemp(8C)

No. of(d)

% weightloss

field release rates(mmol tube21 d21)

NO3 NH4 PO4

Laboratory release rates(mmol tube21 d21)

NO3 NH4 PO4

27 May 9324 Aug 9301 Oct 93

252815

423849

554937

424225

464628

5.55.53.2

48

28

38

26

5.0

5.0

from entering or leaving enclosures during extreme hightides.

The experiment’s two main effects, small predator manip-ulation (0 pinfish 5 2PF and approximately 2–43 meanpinfish density 5 1PF) and nutrient enrichment (ambient 52N and 33 ambient concentrations 5 1N), were tested ina factorial design. Each treatment had six replicates andtreatments were assigned in two rows of 12 plots. To avoidpossible cross-contamination, nutrient treatments wereplaced in the first and last three cages of each row withadditional spacing between the nutrient and nonnutrient cag-es. Preexperiment sampling were done on 26 May 1993 andall treatments were in place by 28 May 1993. The experi-ment was carried out for 176 days with the final samplingon 19 and 20 November 1993.

Nutrient additions—In each of the nutrient enclosures, tenPVC tubes (6 cm diameter 3 30 cm long with twenty 1 cmholes) containing 500 g of OsmocoteTM (N : P molar ratio 58.3) slow release fertilizer were suspended on metal rodsapproximately 10 cm above the bottom within the seagrasscanopy, and evenly spaced throughout the cage. Nutrientswere replaced at approximately six week intervals althoughthe exact time of replacement was dictated by the ambientwater temperature and the results of our laboratory studiesinvolving nutrient dissolution (see below). Nutrient tubeswere cleaned with wire brushes as needed.

In order to characterize the dissolution characteristics ofOsmocotey and to estimate nutrient loading rates in the en-closures, laboratory experiments were conducted to deter-mine Osmocotey dissolution rates at temperatures common-ly encountered during the seagrass growing season (158C,208C, 258C, and 308C) (Fig. 2). Individual PVC tubes con-taining 500 g of Osmocotey were submerged in 4 L of 30PSU seawater in 5 liter glass aquaria mixed with magneticstirrers. Five replicates were used for each temperature treat-ment. Samples were collected on days 0 and 1, and subse-quently at approximately three day intervals over a 10–15day period. These experiments showed the fertilizer releaseto occur in two phases, an initial ‘‘burst’’ phase followed bya constant release rate until the nutrients were exhausted(Fig. 2). For temperatures between 158C and 258C, the con-stant release rates displayed a N : P molar release ratio of17 : 1 in seawater.

Nutrient loading rates were subsequently calculated di-rectly from field measurements. Three times during the ex-periment, 20 nutrient tubes were collected and dried to a


Fig. 3. Nitrate and ammonium concentrations (mM 1 1 SE) bymonth and treatment during 1993. Sampling periods following nu-trient additions are indicated by May B and Aug B. Treatments arerepresented by:2N 5 no nutrients; 1 N 5 nutrient additions;2PF5 no pinfish additions; and 1 PF 5 pinfish additions.

Fig. 4. Phosphate (mM 6 1 SE) and Chl a (mg L21 6 SE)concentrations by month and treatment during 1993. Sampling pe-riods following nutrient additions are indicated by May B and AugB. Treatments are represented by: 2N 5 no nutrients; 1N 5 nu-trient additions; 2PF 5 no pinfish additions; and 1PF 5 pinfishadditions.

constant weight after their deployment. Osmocotey lossrates in the field (g tube21 d21) were calculated based on thechange in Osmocotey weight over time, and N and P deliv-ery rates (mmol tube21 d21) and loading rates (mmol m22

d21) were calculated. Comparisons with the laboratory datashow the two estimates to agree within 610% (Table 1).

Fish Manipulations—Pilot studies showed that our exclo-sure cages effectively excluded large predators (e.g., sharks,red drum, spotted sea trout, and jacks) while allowing small-er invertebrates (e.g., grass shrimp) and benthic fish (e.g.,gobies) access (cf. Leber 1985).

The pinfish, L. rhomboides, dominates the small fish faunain Gulf of Mexico seagrass meadows during spring-fall (Hel-lier 1962; Hansen 1969; Stoner 1982; Huh 1984; Stoner and

Livingston 1984; Livingston 1984) and is the numericallydominant semidemersal fish in St. Joseph Bay (Kip Thomp-son, unpubl. data). Density in seine samples at our study siteranged from 4.57 m22 in May to 0.14 m22 in September and0 m22 during the winter months after young-of-the-year pin-fish had migrated offshore (Kip Thompson, unpubl. data).Similarly, mean pinfish density over an annual cycle in Red-fish Bay, Texas was found to be 2.3 m22 (Huh 1984).

Pinfish undergo ontogenetic changes in feeding behavior(Carr and Adams 1973; Huh and Kitting 1985; Luczkovichand Stellwag 1993) and they have been suggested to playan influential role in controlling both invertebrate and epi-phyte abundance (Stoner 1982; Stoner and Livingston 1984;Huh and Kitting 1985). Based on stomach content analyses,distinctive feeding stages have been characterized, which

1046 Heck et al.

Table 2. Mesograzer categories used in classification of mobileepibiota.

Gastropoda

TurboRissoinaModulusDiastomaCerithiumCrepidulaMitrellaAnachisDentimargo

ChitonIsopodAmphipodPenaeidCaridean (no alpheids)PaguridMajid

vary depending on the area sampled and time of collection(Carr and Adams 1973; Stoner and Livingston 1984; Huhand Kitting 1985). Stoner and Livingston (1984) found fivefeeding stages: (1) planktivore (11–15 mm standard length[SL]); (2) carnivore (16–35 mm SL); (3) omnivore (amphi-pod-dominated) (36–80 mm SL); (4) omnivore (epiphyte-dominated) (81–120 mm SL); and (5) herbivore (.120 mmSL). Livingston (1984) characterizes an omnivorous stage(26–60 mm SL) during which pinfish feed heavily on bothamphipods and epiphytes. With further growth (61–120 mmSL), pinfish become increasingly more herbivorous and fi-nally (.120 mm SL) feed most heavily on seagrasses.

Young-of-the-year pinfish, Lagodon rhomboides, werecaptured by trawling and used to stock 1PF cages at a den-sity of 200 cage21 or 28 m22. The majority of fish rangedfrom 80 mm to 120 mm in initial SL although extremesranging to 180 mm SL were also present. Prior to pinfishadditions all cages were seined to remove fish larger thanthe cage mesh. In the northern Gulf of Mexico, pinfish arepresent at this size (.80 mm SL) by late spring—early sum-mer and could be retained by the mesh enclosures.

Compared with the density of small semidemersal pred-ators at our study site in May, our small predator treatmentscontained approximately 6.23 ambient ‘‘natural’’ densities(28 fish m22). However, because sampling of juvenile fishabundance using seines and trawls typically underestimatesfish density by 30–70% (Kjelson 1977), we estimate that ourinitial stocking density was actually much lower than 6.23‘‘natural’’ pinfish densities. For example, if our estimate of4.5 fish m22 were only 30% of actual densities, the actualabundance would be around 15 m22. If they represented70%, actual densities would be around 6.4 m22. Therefore,we estimate that our study densities ranged from around 1.9–4.43 ‘‘natural densities’’. These enhanced pinfish densitieswere used to simulate what might happen if large predatorswere removed from the system by overharvesting.

Sampling regime—All experimental plots were sampledmonthly from May to November, the period of greatest sea-grass growth in the northern Gulf of Mexico (Iverson andBittaker 1986; Valentine and Heck 1991; 1993). During eachsampling, triplicate water samples were collected from eachenclosure for documentation of inorganic nutrient and water-column Chl a concentrations. To minimize disturbance with-in the enclosures, water samples were collected at the canopyheight from outside each enclosure. A 2-m long aluminumpipe containing Tygon tubing was inserted through the cagemesh and samples were collected using acid-washed 60 mlsyringes. Water samples were placed on ice until (,2 h) theycould be filtered through WhatmanTM GF/C filters and fro-zen in 60 ml plastic bottles. Nutrient analyses were carriedout on samples using standard wet chemical techniques(Alpkem Manual 1988) adapted for use on an Alpkem RFA/2 Nutrient Autoanalyzer. Chl a concentrations were deter-mined using a Turner Designs Model 10 Fluorometer fol-lowing the acidification method of Lorenzen (detailed inStrickland and Parsons 1972).

Seagrass parameters were determined from shoots col-lected in three haphazardly selected 0.01 m2 samples perenclosure. Samples were placed in 5% formalin and stored

for later analyses. In the lab, five randomly selected shootsfrom each sample were used to measure leaf length andwidth and to quantify attached epibionts on the leaves. Epi-phyte species growing on the outside surface of the oldestleaf on each shoot were covered by a 4 3 4 mm grid andthe proportion of grid intersections containing filamentousgreen algae were recorded. (We were especially interested indetermining whether the abundance of filamentous green al-gae, a well-recognized indicator of nutrient enriched condi-tions, would increase in response to nutrient additions). Epi-biont biomass was determined by scraping all sample leaveswith a razor blade and drying to a constant weight at 908C.Samples were then ashed in a muffle furnace at 5008C for 3h and epibiont ash free dry weight (AFDW) was determined.Total seagrass biomass was also determined by drying to aconstant weight at 908C.

Net aboveground primary production was estimated usinga modified blade-marking technique (cf. Dennison 1990).Five randomly selected shoots within 12 cages (3 cages pertreatment) were marked with a probe, identified with aflagged stake and collected within 14 days after marking.Newly produced material below the probe mark and all newblades were dried to a constant weight at 908C to estimatenet aboveground primary production (g DW shoot21 d21).

Invertebrate mesograzers were collected from each enclo-sure using a 0.07 m2 plastic cylinder, whose lower edge wasembedded in the sediment. The macrofauna in the cylinderwere sampled by a gasoline-powered suction pump (cf. Orthand van Montfrans 1987; Williams et al. 1990; Valentine andHeck 1993) and all material was passed through a 0.5 mmcollecting bag where larger motile epibiota were retained.Following collection, samples were sieved on a 0.5 mmmesh screen to remove additional material, placed on ice andfrozen. Animals were identified only to the extent necessaryto be classified trophically (Table 2), according to publishedinformation (cf. Zimmerman et al. 1979; Orth and van Mont-frans 1984; Klumpp et al. 1992, Neckles et al. 1993; Wil-liams and Ruckelshaus 1993). Free-living amphipods, iso-pods, caridean shrimp, and gastropods constituted themajority of mesograzers.

Cage inspection and repair was conducted biweekly, andcage cleaning was conducted monthly to ensure enclosureintegrity. During cage inspections sea urchins, all observedpredatory portunid and xanthid crabs, and other fish species


Tabl

e3.

Com

pari

son

ofes

tim

ated

load

ing

rate

san

dm

easu

red

nitr

ogen

and

phos

phor

usco

ncen

trat

ions

inth

isst

udy

wit

hth

ose

ofpr

evio

usex

peri

men

tal

man

ipul

atio

nsin

seag

rass

syst

ems

and

sele

cted

estu

arie

s.

Fie

ldm

anip

ulat

ions

N-l

oadi

ng(m

mol

m2

2d2

1 )P

-loa

ding

(mm

olm

22

d21 )

Am

bien

t([

N]

mM

)T

reat

men

t([

N]

mM

)A

mbi

ent

([P

]m

M)

Tre

atm

ent

([P

]m

M)

Thi

sst

udy*

Har

lin

and

Tho

rne-

Mil

ler

(198

1)*

Nec

kles

etal

.(1

993)

†W

illi

ams

and

Ruc

kels

haus

(199

3)†

McG

lath

ery

(199

5)

77–1

231,

971–

3,85

7

476–

510

300

5–7

257–

429

34.5

0.3–

5.3

0.4

–2.3

4.0

–10.

83.

6–45

.01.

0–2

.0

0.5–

7.0

1.0

–21.

610

.6–3

7.8

3.6–

45.0

4.0

–4.

5

0–0

.14

0–1

.00.

7–1.

6

,0.

05

0.1–

0.4

0.1–

5.6

1.8–

3.4

0.10

Sel

ecte

des

tuar

ies‡

Apa

lach

icol

aB

ayC

hesa

peak

eB

ayD

elaw

are

Bay

Nar

raga

nset

tB

ayT

ham

esE

stua

ry,

U.K

.

1.9

19.2 2.7

87.7

0.11

1.60

0.22

8.70

5.4

20.0

110.

08.

1

0.2

0.5

1.2

1.9

*D

INad

ded

asN

O3

and

NH

4.†

DIN

adde

das

NH

4on

ly.

‡E

stim

ated

from

Nix

on(1

986)

.

larger than 2 cm in body width or height were removed byhand or by spearing.

After the final sampling, all fish in each cage were col-lected, counted and identified. Up to 20 pinfish per cage weremeasured (TL and SL in mm) and weighed (wet weight).Fultons condition factor was calculated (K 5 weight 3length23) (Lagler et al. 1962) and used to investigate treat-ment differences in pinfish condition.

Statistical analyses—Two-way repeated measures ANO-VA was used to analyze the effects of small predator ma-nipulation and nutrient enrichment on monthly measure-ments of seagrass parameters, mesograzer abundance,epibiont DW, AFDW, and percent occurrence by taxonomiccategory, inorganic nutrients, and Chl a. Nutrient concentra-tions, final fish counts, and condition factors were also an-alyzed using two-way ANOVAs. Scheffe’s multiple com-parison procedure was used when P , 0.05 for main effectsand interactions were not significant. Data were transformedwhen necessary to meet assumptions of the ANOVAs.

Results

Water column nutrients—While nutrient concentrationsvaried somewhat during the experiment as a result of naturalinput and wind-mixing, on average, 1N treatments achieved33 ambient nutrient levels observed in 2N plots (Figs. 3,4).Both nitrogen and phosphate levels generally remained ele-vated in the nutrient addition treatments throughout the du-ration of the experiment (Figs. 3,4). High concentrationswere observed following the addition of replacement fertil-izer tubes, as documented in May and August (Figs. 3,4);however, these periods were expected to be brief based onlaboratory dissolution results.

Initial ambient water samples collected in May were notsignificantly different in nitrate, nitrite, ammonium, phos-phate, silicate, or Chl a levels among treatments. Throughoutthe experimental duration, nitrite and silicate levels werevery low and similar between treatments and within months,thus we do not present the results for these nutrients. Chl aconcentrations were also low (,2 mg L21) and showed notreatment effects except immediately after the initial nutrientaddition in May (Fig. 4). We believe that this response wasan artifact caused by the physical dislodging of epiphytesduring initiation of the experiment, not phytoplanktongrowth.

Over all months, nitrate concentrations were greatest inthe nutrient addition treatments. Nitrate levels in the 2Ntreatments ranged from 0.02 mM (2N2PF) in August to0.56 mM (2N1PF) in July (Fig. 3). In the nutrient additioncages, nitrate levels ranged from 0.53 mM (1N2PF,1N1PF) in August to 1.39 mM (1N2PF) in September.Excluding the initial sampling date, nitrate concentrationswere significantly higher in the 1N treatments comparedwith the 2N treatments.

Ammonium levels in the 2N treatments were lowest inAugust (0.21 mM1 : 2 2N1PF) and highest during July(2.03 mM1 : 22N1PF) (Fig. 3). Ammonium concentrationsin nutrient enriched treatments ranged from 0.69 mM(1N1PF) in August to 3.14 mM (1N2PF) in September.

1048 Heck et al.

Fig. 5. Ratio of epibiont dry weight and epibiont ash free dryweight (g 0.01 m22) to leaf dry weight (g 0.01 m22) (6SD) bymonth and treatment during 1993. Treatments are represented by:2N 5 no nutrients; 1N 5 nutrient additions; 2PF 5 no pinfishadditions; and 1PF 5 pinfish additions. May samples were takenbefore the experiments were begun.

Fig. 6. Proportion of leaves covered by filamentous green algaeby month and treatment during 1993. Treatments are representedby: 2N 5 no nutrients; 1N 5 nutrient additions; 2PF 5 no pinfishadditions; and 1PF 5 pinfish additions. May samples were takenbefore the experiments were begun.

Significant nutrient effects occurred for all sampling datesexcept July and October (Fig. 3).

Phosphate concentrations ranged from undetectable (Sep-tember: 2N2PF) to 0.29 mM (July: 2N2PF) in the 2Ntreatments (Fig. 4). Phosphate values were highest in July(0.27 mM1 : 21N-PF) and lowest in November (0.02 mM1 :21N1PF). Phosphate concentrations were significantlygreater within nutrient enriched cages compared with theambient treatments in June (F2,23 5 11.84, P 5 0.003), Sep-tember (F2,23 5 13.38, P 5 0.002) and November (F2,23 515.63, P , 0.001) (Fig. 4).

During the May ‘‘initial burst’’, significant nutrient treat-ment effects were documented for nitrate (F2,23 5 41.46,P , 0.001), ammonium (F2,23 5 14.95, P , 0.001), phos-phate (F2,23 5 25.71, P , 0.001), and Chl a (F2,23 5 8.46,P 5 0.009). Nutrient effects were also observed in the Au-

gust burst for nitrate (F2,23 5 29.25, P , 0.001), ammonium(F2,23 5 9.64, P 5 0.006) and phosphate (F2,23 5 13.48, P5 0.002). Nutrient enrichment treatments had elevated levelsof nitrate, ammonium, and phosphate in May and Augustcompared with the 2N treatments.

Epibionts—Epibiont DW and AFDW/leaf DW was usu-ally greater in cages without pinfish than in those with pin-fish additions, especially during the summer months (Fig. 5),and these ratios were significantly lower in pinfish additiontreatments (Table 4). Month, as well as interactions betweenfish and month, were also significant in both cases (Table4).

Overall, significant between subjects nutrient and fish ef-fects were documented for the percentage of filamentousgreen algal cover, with significantly more filamentous greenalgae in 1N treatments, and significantly less algae in 1PFtreatments (Table 4; Fig. 6). Month, two-way (nutrient 3month and fish 3 month) and three-way interaction terms(nutrient 3 fish 3 month) were also significant within sub-jects.

Mesograzers—Free-living amphipods, isopods, carideanshrimp, hermit crabs, and gastropods constituted the majorityof mesograzers at our study site. As expected, pinfish re-duced mesograzer densities during most months (Fig. 7), re-sulting in significant fish, but not nutrient, effects (Table 4).There were also significant monthly differences that reflectedusual seasonal variations in population size (Table 4).

Seagrass—Leaf length varied seasonally, with greatestmean lengths of 25–30 cm occurring in July–September(Fig. 8). Leaves collected from pinfish cages were signifi-cantly longer than leaves from the no pinfish treatments (Ta-ble 4). Increased nutrient levels did not significantly affect


leaf length although there were significant seasonal effectson leaf length (Table 4).

Seagrass leaf width was not significantly affected by eithernutrient or pinfish treatments (Table 4), but did show sig-nificant seasonal variability (Table 4; Fig. 8).

Seagrass biomass peaked during July, with mean valuesranging between approximately 95 and 135 g dry wt m22.Significant pinfish effects were observed, with seagrass bio-mass usually greatest in pinfish treatments (Table 4; Fig. 9).There were significant seasonal effects on seagrass biomassas well as a fish 3 month interaction (Table 4). Seagrassshoot number responded similarly, with significantly en-hanced shoot densities present in pinfish treatment, espe-cially during the August–November period (Table 4; Fig. 10)and significant seasonal variability (Table 4). Although sea-grass production was not significantly affected by either nu-trient or fish treatments (Table 4), owing to large variability,the greatest production rates generally occurred in the 1PFtreatments from August through October (Fig. 11). A sig-nificant seasonal effect was also noted (Table 4).

Experiment termination—A final sampling was conductedin November; in addition, all cages were seined to quantifyfinal fish densities. Initially, L. rhomboides was stocked at 200fish per cage (1PF treatments). Seven months later in Novem-ber, pinfish numbers had declined in all pinfish enclosures (pre-sumably because of escapes), so that the 1PF treatments con-tained approximately 33 the density of nonpinfish enclosures(mean of 66.0 cage21 or 9.4 m22 in 1PF enclosures, comparedwith a mean of 21.8 cage21 or 3.1 m22) in 2PF treatments(which often contained unwanted immigrants) (Fig. 12). Thisdifference was significant (F2,23 5 24.23, P , 0.0001). Addi-tional immigrant fish species removed from cages includedspot, cowfish, pigfish, cubbyu, gag grouper, toadfish, pipefish,filefish, sand perch, mullet, goby, red snapper, sheephead, searobin, seahorse, flounder, and speckled trout. These additionalspecies made up a small fraction of the total fish number bycage and treatment, and never exceeded a mean value of morethan 2 cage21 (Fig. 12).

Pinfish in the enclosures at the end of the experimentranged from 50 mm to 200 mm in total length (TL), withthe majority in the 60–100 mm TL range (Fig. 13). Condi-tion factors (K 5 weight 3 length23) (Lagler et al. 1962)were affected by pinfish density (F2,23 5 9.22, P 5 0.0065)but not by nutrients (Fig. 14). Pinfish from 1PF cages hada significantly lower condition index (mean 5 0.025) com-pared with intruding pinfish found in the no 2PF treatments(mean 5 0.027) (Fig. 14), although we ascribe little biolog-ical importance to this small difference in condition index.

Discussion

Duarte (1995) has summarized the conventional wisdomregarding the effects of nutrient enrichment on SAV, statingthat increasing nutrient supplies lead to the overgrowth ofseagrasses by fast-growing algae and the eventual disap-pearance of seagrasses from eutrophic systems. This is theexplanation most often proposed to account for the loss ofseagrasses in North America (Orth and Moore 1983; Neun-dorfer and Kemp 1993; Short et al. 1995; Tomasko et al.

1996), Europe (Giesen et al. 1990; Den Hartog 1994) andAustralia (Cambridge and McComb 1984). However, noneof the studies cited above evaluated the ability of epiphytegrazers (mesograzers) to control epiphyte biomass on sea-grass leaves. This is relevant because numerous studies havedemonstrated that small crustacean and gastropod grazerscan control epiphyte biomass. For example, van Montfranset al. (1982), Robertson and Mann (1982), and Howard andShort (1986) demonstrated that small gastropods (e.g., Bit-tium varians) could regulate epiphyte biomass in laboratorymicrocosms. Similarly, Caine (1980) and Howard (1982)showed that amphipods could control epiphyte biomass, withCaine (1980) reporting differences of up to 400% in epiphytebiomass between grazed and ungrazed eelgrass leaves. Inaddition, Borum (1987) demonstrated that a mixed group ofamphipods, isopods, and gastropods at field densities keptepiphyte biomass on eelgrass shoots at levels only 10% thosein ungrazed treatments. Reviews of the many studies thatdemonstrate the remarkable degree to which mesograzerscan control epiphyte abundance on macrophytes are provid-ed by van Montfrans et al. (1984), Brawley (1992), and Jer-nakoff et al. (1996). On balance, these reviews suggest thatwhen grazers are present, the stimulatory effects of increasednutrient loading on epiphyte abundance are greatly reduced.

Recently, several nutrient enrichment experiments withgrazers present have found only partial support for the sim-ple nutrient enrichment hypothesis cited by Duarte (1995).For example, Neckles et al. (1993) found that epiphyte graz-ing by amphipods seasonally prevented the overgrowth ofeelgrass by algae in nutrient-enriched mesocosms, and con-cluded that the effects of grazers were stronger than thoseof nutrients. Williams and Ruckelshaus (1993) found thatisopod grazing reduced epiphyte biomass by one-third, whilein the absence of grazing nutrient enrichment led to in-creased epiphyte biomass that negatively affected eelgrassgrowth. They concluded that epiphytes have the potential tocontrol eelgrass growth only when small grazers are absent,nitrogen in the water column is abundant (.15 mM DIN;dissolved inorganic nitrogen) and when temperatures aresuboptimal for eelgrass growth. In the Netherlands, Phillipart(1995) used field enclosure experiments to show that gastro-pod (Hydrobia ulvae) grazing on epiphytes led to enhancedeelgrass density and biomass. Phillipart (1995) further sug-gested that eelgrass declines in the Wadden Sea during thepast 25–30 yrs, which have been attributed to the effects ofeutrophication, may have instead been initiated when Hy-drobia abundances declined precipitously in the early 1970’s(Phillipart 1995). Lin et al. (1996) assessed the effects ofnutrient enrichment on laboratory mesocosms of eelgrasscontaining mesograzers as well as faunal components fromall trophic levels. They found that epiphyte biomass was nota good indicator of nutrient loading in shallow coastal la-goons, and that epiphyte responses to nutrient enrichmentwere surprisingly complex. Of primary importance, theyconcluded, were system level faunal interactions that pro-duced results that differed from previous studies which didnot include complex plant–animal interactions (Lin et al.1996). In aggregate, the results from these studies showedthat mesograzers frequently control the abundance of epi-phytes, even in enriched conditions.

1050 Heck et al.

Table 4. Univariate results of repeated measures ANOVAs.

Source of variation DF Type III SS Mean square F value Pr . F

Epibiont DW/Seagrass DWBetween subjects

NutrientsFishNutrients 3 Fish

111

0.005136990.439622890.05952768

0.005136990.439622860.05952768

0.1916.00

2.17

0.66620.00010.1436

Within subjectsDateNutrients 3 DateFish 3 DateNutrients 3 Fish 3 Date

6665

5.587091390.059527681.517282800.14442732

0.931181900.009919070.252880470.02888546

33.900.369.211.05

0.00010.90220.00010.3908

Epibiont AFDW g Seagrass DW21

Between subjectsNutrientsFishNutrients 3 Fish

111

0.015088620.331370710.14739596

0.015088620.331370710.04739596

0.8418.37

2.63

0.36220.00010.1076


6665

3.623849950.066742791.008290830.08765056

0.603974990.011123800.168048470.01753011

33.480.629.320.97

0.00010.71660.00010.4379

Percent leaf cover by green filamentous algaeBetween subjects


111

0.081237530.018839820.00033254

0.081237530.018839820.00033254

35.418.210.14

0.00010.00480.7040


6666

1.207291720.191184390.040889640.05277404

0.201215290.031864070.006814940.00879567

87.7013.89

2.973.83

0.00010.00010.00920.0014

Number of mesograzers g Seagrass DW21


111

26103.8707359639.2452

468.5076

26103.8707359639.2452

468.5076

1.8325.15

0.03

0.17890.00010.8566


6666

517813.211375148.406544863.216313243.8359

86302.201912524.73447477.20272207.3060

6.030.880.520.15

0.00010.51450.79030.9879

Mean leaf length (cm)Between subjects


111

11.07846189.29133135.087021

11.07846189.29133135.087021

0.786.292.47

0.37870.01330.1183


6666

4067.55794260.76174196.01400930.388498

677.92632410.12695716.002335

5.064750

47.720.711.130.36

0.00010.63980.35010.9051

Mean leaf width (mm)Between subjects


111

0.090807950.016402030.00064042

0.090807950.016402030.00064042

3.010.510.02

0.08470.46180.8843


Table 4. Continued.

Source of variation DF Type III SS Mean square F value Pr . F


6666

1.751818800.131032200.041257160.02717141

0.292469800.021838700.006876190.00452857

9.710.720.230.15

0.00010.63030.96690.9887

Total leaf dry weight (g DW m2)Between subjects


111

0.011619562.809430810.26476576

0.011619562.809430810.26476576

0.1024.67

2.33

0.74990.00010.1295


6666

40.210160760.453266202.650586660.19895515

6.701393460.075544370.441764440.03315919

58.860.663.880.29

0.00010.67920.00130.9403

Number of shoots 0.01 m2-1


111

3.379489878.857870370.51580189

3.379489878.857870370.51580189

2.436.370.37

0.12240.01330.5440


4444

68.932112462.423450575.605126190.39948103

17.233028110.605862641.401281550.9987026

12.390.441.010.07

0.00010.78270.40770.9905

Production (g DW shoot21 day21)Between subjects


111

0.000000100.000008800.00000000

0.000000100.000008800.00000000

0.0330.6

0.00

0.85410.08650.9945

Within subjectsDateNutrients 3 DateFish 3 DateNutrient 3 Fish 3 Date

5555

0.000230910.000007630.000019030.00000239

0.000046180.000001530.000003810.00000048

16.000.531.320.17

0.00010.75180.27000.9738

In our experiments, we found that nutrient enrichment hadno significant effect on epiphyte biomass, or the production,leaf length, shoot density, or biomass of T. testudinum. Infact, the only significant nutrient effect was on the increasedcover of filamentous green algae on seagrass leaves. In con-trast, manipulation of pinfish densities resulted in significanteffects on mesograzer density, epiphyte biomass, and theproduction, leaf length and shoot density of Thalassia. Theseresults are clearly not consistent with the simple paradigmof nutrient-enrichment based seagrass decline summarizedby Duarte (1995) and require close examination.

Initially, it is important to assess the level of nutrient en-richment in the 1N treatments. We estimate (based on datafrom both 158C and 258C; Fig. 2) that our 1N treatmentsreceived 77–123 mmol DIN m22 d21, and 5–7 mmol P m22

d21. These daily rates translate to annual rates of 28–45 molDIN m22 yr21 and 1.8–2.6 mol PO4 m22 yr21 and are gen-erally much higher than rates estimated for major estuariesof the world, and within the range of those achieved in otherseagrass nutrient enrichment studies (Table 3). While someadvection of nutrients from our experimental cages undoubt-

edly occurred, there are a number of reasons why we believethat our nutrient enrichments were quite effective: (1) St.Joseph Bay is a low energy environment, and current move-ment is negligible in the southern portion of the Bay (Stew-art and Gorsline 1962), where our study site was located.This is supported by the low average current velocities wemeasured (around 2.8 cm sec21), indicating that large scaleadvection of dissolved nutrients is unlikely; (2) placing thenutrient delivery tubes within the seagrass canopy ensuredthat nutrient release occurred in an area of very low flow(Fonseca et al. 1982; Ackerman and Okubo 1993), and im-mediately adjacent to and accessible by seagrass algal epi-phytes; (3) we measurably increased water-column nutrientconcentrations; (4) there was an increase in the proportionof filamentous green epiphytes in the nutrient treatments(Fig. 6), indicating microalgal responses to enhanced nutri-ent loading; and (5) an increase of nearly 25% in N concen-trations was recorded in Thalassia leaves in 1N treatments(Lores, USEPA Gulf Breeze Laboratory unpubl. data), in-dicating that increased concentration of N was achieved andavailable in our 1N treatments. And, subsequent to our ex-

1052 Heck et al.

Fig. 7. Ratio of grazer number to leaf dry weight (g) (6SD) bymonth and treatment during 1993. Treatments are represented by:2N 5 no nutrients; 1N 5 nutrient additions; 2PF 5 no pinfishadditions; and 1PF 5 pinfish additions. May samples were takenbefore the experiments were begun.

Fig. 8. Mean leaf length and width (cm) (6SD) by month andtreatment during 1993. Treatments are represented by: 2N 5 nonutrients; 1N 5 nutrient additions; 2PF 5 no pinfish additions;and 1PF 5 pinfish additions. May samples were taken before theexperiments were begun.

periments, Thomas et al. (unpubl. data) have carried out insitu flume studies at the sampling site that have shown veryhigh uptake of water-column nutrients by the seagrass–epi-phyte complex over very short (5–6 min) time periods.

An important characteristic of our 1N treatment was thatdespite high nutrient loading, nutrient concentrations wereseldom elevated above 10 mM DIN and 0.3 mM PO4. Thus,some potential consequences of nutrient enrichment, such asenhanced growth of attached algae with high half-saturationcoefficients for nutrient uptake or phytoplankton blooms, didnot occur. However, our data clearly support the fact thatdespite significant nutrient loading the expected large in-creases in the biomass of epiphytes and associated seagrassdecline did not occur.

In contrast to the 1N treatments, the 1PF (elevated pin-fish density) treatments displayed numerous significant ef-fects. Significant reductions in mesograzer densities were ob-served in 1PF (Fig. 6), yet there was no evidence of theexpected epiphyte proliferation, even in the 1N enclosures(Fig. 5). Instead, increased pinfish numbers led to reductionsin the biomass of epiphytes. We believe that two factorsaccount for this unanticipated result. First, pinfish consumeincreasing amounts of epiphytes as they grow (Livingston1982; Stoner and Livingston 1984; Luczkovich and Stellwag1993), and they were likely to have been consuming sub-stantial amounts of epiphytes. Second, mesograzer density,although significantly reduced in pinfish treatments, was stillvery high (e.g., ;3 3 103 individuals m22 in July, and 2.73 103 m22 in August, as obtained by multiplying numbersof mesograzers per g DW [Fig. 6] by the mean seagrass DWin pinfish treatments [Fig. 8]). These numbers are above thehighest density recorded (2.1 3 103 m22) by Nelson (1980)in a survey of amphipod densities in North American sea-

grass meadows from Nova Scotia to Florida, and are about75% of the mean density of macrobenthic species recordedfrom densely vegetated turtlegrass meadows in ApalacheeBay, Florida by Stoner (1982). They are also above the rangeof mesograzer densities (1–2 3 103 m22) found by Howard(1982) and Howard and Short (1986) to produce significantepiphyte reductions in mesocosm experiments and substan-tially greater than the isopod densities (;100 m22) reportedby Williams and Ruckelshaus (1993). Neckles et al. (1993)observed significant epiphyte reductions at amphipod den-sities of ;4,800–11,400 m22. Based on the weight of theabove evidence, it appears that the relatively high mesogra-zer numbers still remaining, combined with the high pinfishdensities, prevented the proliferation of epiphytes in our el-evated pinfish treatments, despite the high loading rates inour 1N treatments.

We can also ask why our elevated pinfish densities didnot drive mesograzer densities below the relatively high


Fig. 9. Mean leaf dry weight (g m22) (6SD) by month andtreatment during 1993. Treatments are represented by: 2N 5 nonutrients; 1N 5 nutrient additions; 2PF 5 no pinfish additions;and 1PF 5 pinfish additions. May samples were taken before theexperiments were begun.

Fig. 10. Mean shoot density (number m22) (6SD) by monthand treatment during 1993. Treatments are represented by: 2N 5no nutrients; 1N 5 nutrient additions; 2PF 5 no pinfish additions;and 1PF 5 pinfish additions.

Fig. 11. Mean leaf production 103 g d21 shoot21 (6SD) by dateand treatment during 1993. Treatments are represented by: 2N 5no nutrients; 1N 5 nutrient additions; 2PF 5 no pinfish additions;and 1PF 5 pinfish additions..

numbers of 2,5001 individuals m22. We think this is ex-plained by the high standing crop of turtlegrass in our treat-ments. Stoner (1982) found that 177.6 g dry wt m22 ofturtlegrass reduced pinfish foraging success on seagrass-as-sociated amphipods to about 20% of that on bare sand sub-strate, while biomasses of 88 and 22 g dry wt m22 reducedforaging success to around 40 and 60%, respectively, of thaton unvegetated substrate. Turtlegrass standing crops in our1PF treatments ranged from 30 g (in November) to 130 gdry wt m22 (in July), but from May to July were never below60 g dry wt m22. Therefore, even though pinfish densities

were substantially elevated in our treatments, the high bio-mass of seagrass still should have provided sufficient pro-tection for prey to allow the relatively high densities of me-sograzers to persist.

Reduced epiphyte biomass produced positive effects onseagrass biomass, and to a lesser extent on shoot density andshoot-specific productivity. This is similar to the results ofboth Neckles et al. (1993) and Williams and Ruckelshaus(1993). The simple explanation for this is that seagrassleaves receive more light and grow more rapidly when lessheavily fouled by epiphytes. Therefore, whenever epiphytegrazing is sufficient to control epiphyte populations, sea-grasses appear to benefit. This is clear evidence of important‘‘top-down’’ effects in these seagrass ecosystems, though wedid not observe a simple alternation of high and low biomassbetween successive trophic levels as seen in simple trophiccascades. Instead, pinfish, while substantially reducing me-sograzers, also fed on algal epiphytes, which partially pre-vented the release of epiphytes from mesograzer control, de-spite high rates of nutrient enrichment. In addition, the highseagrass density also allowed high numbers of mesograzersto persist even in the face of strong predation pressure bypinfish. Effectively, the omnivorous pinfish, together withthe remaining numbers of mesograzers, prevented a simpletrophic cascade from occurring. This is, at least in part, whatwas predicted by Strong (1992) and Polis and Strong (1996),who have argued that trophic cascades are uncommon exceptin simple food chains because of the reticulate nature of mostfood webs.

If this interpretation of our results is correct, it has impli-cations for better understanding the effects of nutrient en-richment on seagrass habitats. In sparsely vegetated mead-ows (e.g., those at depths where light may be limiting toseagrass biomass accumulation, or in unfavorable salinitiesor temperature regimes) subjected to eutrophication, the low

1054 Heck et al.

Fig. 12. Mean number of pinfish and other fish species by treat-ment seined from cages in November 1993. Treatments are repre-sented by: 2N 5 no nutrients; 1N 5 nutrient additions; 2PF 5no pinfish additions; and 1PF 5 pinfish additions.

Fig. 13. Total length data (mm) of pinfish seined from all cagesin November 1993.

Fig. 14. Fulton condition index (6SD) by treatment for a sub-sample of pinfish (#20) seined from cages in November 1993. Thecondition index was determined using the equation: Condition index5 weight 3 length23. Treatments are represented by: 2N 5 nonutrients; 1N 5 nutrient additions; 2PF 5 no pinfish additions;and 1PF 5 pinfish additions.

seagrass biomass is unlikely to afford adequate protectionfor mesograzer numbers to remain high enough to controlthe growth of epiphytes as they respond to eutrophication.Epiphyte overgrowth of seagrasses should therefore proceedrapidly in eutrophic coastal areas which have also lost large,piscivorous predators to overfishing, and whose populationsof mesograzer-consuming small fishes are consequently atvery high levels. One could also expect the susceptibility ofseagrass meadows to eutrophication to vary annually as ei-ther the density of seagrasses fluctuates in response to chang-ing environmental factors (e.g., varying light levels due toincreased runoff or storm activity, or varying temperaturesor salinities), or as small fish populations vary in responseto changes in year-class strength. Both of these effects wouldresult in changes in the number of mesograzers present. Inaddition, environmental factors that reduce seagrass growthrates (e.g., increased turbidity, stressful temperature, or sa-linity) could also increase the meadow’s susceptibility to eu-trophication by allowing epiphytes more time to accumulateon slowly growing seagrass leaves.

This scenario elaborates on the conditions proposed byWilliams and Ruckelshaus (1993) that can lead to epiphyteproliferation and subsequent seagrass decline: unfavorabletemperatures for seagrass growth, high availability of DIN,and few grazers. We consider the role of seagrass biomassto be critical in determining the size of mesograzer popu-lations, especially where the larger predators of small fishesare not abundant. And we believe that low seagrass biomasscould often be responsible for keeping mesograzer abun-

dances too low to prevent the overgrowth of seagrass leavesby epiphytes when eutrophication accelerates.

Overall, we believe that our data: (1) support the resultsof previous studies that manipulated both nutrient supply andgrazing activity, in showing that grazers frequently can con-trol the abundance of epiphytes; and (2) indicate that it isunlikely we can gain a full understanding of the consequenc-es of nutrient enrichment for seagrass ecosystems without


understanding what controls the population fluctuations ofmesograzers. The potential importance of grazers in deter-mining the abundance of seagrass epiphytes was pointed outsome time ago (e.g., Orth and van Montfrans 1984; vanMontfrans et al. 1984), although recent investigations havefocused more on the effects of nutrient supply in controllingepiphyte abundance, with a smaller role described for theeffects of grazing (e.g., Duarte 1995).

Our results also indicate important ‘‘top-down’’ effects indetermining the structure of seagrass communities. Althoughwe did not observe simple trophic cascades, presumably be-cause pinfish feed at several different trophic levels, it isquite clear that the abundance of small fishes can have im-portant consequences for small plants and animals as wellas seagrasses themselves. Finally, the importance of seagrassmeadows in controlling secondary productivities in coastalsystems strongly suggests that additional tests of the hy-potheses laid out above are necessary. It appears that therole of mesograzers in controlling the effects of eutrophi-cation in seagrass meadows has been underestimated, and inan effort to preserve the health of seagrass ecosystems, itshould be well worth the effort to understand mesograzereffects more completely.

References

ACKERMAN, J. D., AND A. OKUBO. 1993. Reduced mixing in a ma-rine macrophyte canopy. Func. Ecol. 7: 305–309.

ALPKEM MANUAL. 1988. Alpkem Corporation, Clakamas, Oregon.BORUM, J. 1987. Dynamics of epiphyton on eelgrass (Zostera ma-

rina) leaves: Relative roles of algal growth, herbivory and sub-stratum turnover. Limnol. Oceangr. 32: 986–992.

BRAWLEY, S. H. 1992. Mesoherbivores, p. 235–263. In D. M. John,S. J. Hakins, and J. H. Price [eds.], Plant-animals interactionsin marine benthos. Clarendon Press.

BRONMARK, C. 1985. Interactions between macrophytes, epiphytesand herbivores: An experimental approach. Oikos 45: 26–30.

, AND S. E. B. WEISNER. 1996. Decoupling of cascadingtrophic interactions in a freshwater benthic food chain. Oec-ologia 108: 534–541.

BROOKS, J. L., AND S. I. DODSON. 1965. Predation, body size andcomposition of plankton. Science 150: 28–35.

CAINE, E. A. 1980. Ecology of two littoral species of caprellid am-phipods (Crustacea) from Washington, USA. Mar. Biol. 56:327–335.

CAMBRIDGE, M. L., AND A. J. MCCOMB. 1984. The loss of sea-grasses in Cockburn Sound, Western Australia. I. The timecourse and magnitude of seagrass decline in relation to indus-trial development. Aquatic Bot. 20: 229–243.

CARPENTER, S. R., AND D. M. LODGE. 1986. Effects of submersedmacrophytes on ecosystem processes. Aquat. Bot. 26: 341–370.

, J. F. KITCHELL, AND J. R. HODGSON. 1985. Cascading tro-phic interactions and lake productivity. Bioscience 35: 634–639.

, AND OTHERS. 1987. Regulation of lake primary productiv-ity by food-web structure. Ecol. 68: 1863–1867.

CARR, W. E. S., AND C. A. ADAMS. 1973. Food habits of juvenilemarine fishes occupying seagrass beds in the estuarine zonenear Crystal River, Florida. Trans. Amer. Fish Soc. 102: 511–540.

COEN, L. C., K. L. HECK, JR., AND L. G. ABLE. 1981. Experiments

on competition and predation among shrimp of seagrass mead-ows. Ecol. 62: 1484–1493.

DAYTON, P. K. 1971. Competition, disturbance and community or-ganization: The provision and subsequent utilization of spacein a rocky intertidal community. Ecol. Mono. 41: 351–389.

DEN HARTOG, C. 1994. Suffocation of a littoral Zostera bed byEnteromorpha radiata. Aquat. Bot. 47: 21–28.

DENNISON, W. C. 1990. Leaf production, p. 77–85. In R. C. Phillips,and C. P. McRoy [eds.], Seagrass research methods. UNESCO.

DUARTE, C. M. 1995. Submerged aquatic vegetation in relation todifferent nutrient regimes. Ophelia 41: 37–112.

DUGGINS, D. O. 1980. Kelp beds and sea otters: An experimentalapproach. Ecology 61: 447–453.

ESTES, J. A., AND J. F. PALMISANO. 1974. Sea otters: Their role instructuring nearshore communities. Science 185: 1058–1060.

FONSECA, M. S., J. S. FISHER, J. C. ZIEMAN, AND G. W. THAYER.1982. Influence of the seagrass, Zostera marina L., on currentflow. Estuar. Coast. Shelf Sci. 15: 351–364.

GIESEN, W. B. J. T., M. M VAN KATWIJK, AND C. DEN HARTOG.1990. Eelgrass condition and turbidity in the Dutch WaddenSea. Aquatic Bot. 37: 71–85.

HANSEN, D. J. 1969. Food, growth, migration, reproduction andabundance of pinfish Lagodon rhomboides, and croaker Mi-cropogon undulatus near Pensacola, Florida 1963–1965. Fish.Bull. 68: 135–146.

HARLIN, M. M., AND B. THORNE-MILLER. 1982. Seagrass-sedimentdynamics of a flood-tidal delta in Rhode Island. Aquat. Bot.14: 127–138.

HECK, K. L., JR., AND L. B. CROWDER. 1991. Habitat structure andpredator–prey interaction in vegetated aquatic systems, p. 281–299. In S. S. Bell, E. D. McCoy, and H. R. Mushinsky [eds.],Habitat complexity: The physical arrangement of objects inspace. Chapman and Hall.

, K. W. ABLE, C. T. ROMAN, AND M. P. FAHAY. 1995. Com-position, abundance, biomass and production of macrofauna ina New England estuary: Comparisons among eelgrass mead-ows and other nursery habitats. Estuaries 18: 379–389.

HELLIER, J. R., JR. 1962. Fish production and biomass studies inrelation to photosynthesis in the Laguna Madre of Texas. Contr.Mar. Sci. 8: 1–22.

HOOTSMANS, M. J. M., AND J. E. VERMAAT. 1985. The effect ofperiphyton-grazing by three epifaunal species on the growth ofZostera marina L. under experimental conditions. Aquat. Bot.22: 83–88.

HOWARD, R. K. 1982. Impact of feeding activities of epibenthicamphipods on surface-fouling of eelgrass leaves. Aquat. Bot.14: 91–87.

, AND F. T. SHORT. 1986. Seagrass growth and survivorshipunder the influence of epiphyte grazers. Aq. Bot. 24: 287–302.

HUGHES, T. P. 1994. Catastrophes, phase shifts, and large scale deg-radation of a Caribbean coral reef. Science 265: 1547–1551.

HUH, S-H. 1984. Seasonal variations in populations of small fishesconcentrated in shoalgrass and turtlegrass meadows. J. Ocean-ological Soc. Korea 19: 44–55.

, AND C. L. KITTING. 1985. Trophic relationships amongconcentrated populations of small fishes in seagrass meadows.J. Exp. Mar. Biol. Ecol. 92: 29–43.

IVERSON, R. L., AND H. F. BITTAKER. 1986. Seagrass distributionand abundance in the eastern Gulf of Mexico water. Est. Coast.Shelf Sci. 22: 577–602.

JERNAKOFF, P., A. BREARLEY, AND J. NIELSEN. 1996. Factors af-fecting grazer-epiphyte interactions in temperate seagrassmeadows. Oceanogr. Mar. Biol. (Ann. Rev.) 34: 109–162.

KAISER, S. C. 1989. The growth responses of eelgrass, Zostera ma-rina, to epiphyte fouling and herbivore grazing in a New Eng-land estuary. M. S. Thesis, Rutgers Univ., New Brunswick.

1056 Heck et al.

KJELSON, M. A. 1977. Estimating the size of juvenile fish popula-tions in southeastern coastal-plain estuaries, p. 71–90. In W.Van Winkle [ed.], Proceedings of the Conference on Assessingthe Effects of Power-Plant-Induced Mortality on Fish Popula-tions. Pergammon Press.

KLUMPP, D. W., J. S. SALITA-ESPINOSA, AND M. D. FORTES. 1992.The role of epiphytic periphyton and macroinvertebrate grazersin the trophic flux of a tropical seagrass community. Aq. Bot.43: 327–349.

LAGLER, K. R., J. E. BARDACH, AND R. R. MILLER. 1962. Ichthy-ology. John Wiley and Sons.

LEBER, K. M. 1985. The influence of predatory decapods, refuge,and microhabitat selection on seagrass communities. Ecol. 66:1951–1964.

LIN, H. J., S. W. NIXON, D. I. TAYLOR, S. L. GRANGER, AND B. A.BUCKLEY. 1996. Responses of epiphytes on eelgrass, Zosteramarina L., to separate and combined nitrogen and phosphorusenrichment. J. Exp. Mar. Ecol. 52: 243–248.

LIVINGSTON, R. J. 1982. Tropic organization of fishes in a coastalseagrass system. Mar. Ecol. Prog. Ser. 7: 1–12.

. 1984. Trophic responses of fishes to habitat variability incoastal seagrass systems. Ecol. 65: 1258–1275.

LODGE, D. M., K. K. KERSHUER, J. E. ALOI, AND A. P. COVICH.1994. Effects of an omnivorous crayfish (Orconectes rusticus)on a freshwater littoral food web. Ecol. 75: 1265–1281.

LUCZKOVICH, J. J., AND E. J. STELLWAG. 1993. Isolation of cellu-lolytic microbes from the intestinal tract of the pinfish, Lago-don rhomboides: Size-related changes in diet and microbialabundance. Mar. Biol. 116: 381–388.

MAIN, K. L. 1985. The influence of prey identity and size selectionof prey by two marine fishes. J. Exp. Mar. Biol. Ecol. 98: 145–152.

. 1987. Predator avoidance in seagrass meadows: Prey be-havior microhabitat selection and cryptic coloration. Ecol. 68:170–180.

MARTIN, T. H., L. B. CROWDER, C. DUMAS, AND J. M. BURKHOLD-ER. 1992. Indirect effects of fish on macrophytes in BaysMountain Lake: Evidence for a littoral trophic cascade. Oec-ologia 89: 476–481.

MCGLATHERY, K. J. 1995. Nutrient and grazing influences on asubtropical seagrass community. Mar. Ecol. Prog. Ser. 122:239–252.

MITTELBACH, G. G. 1984. Predation and resource partitioning intwo sunfishes (Centrarchidae). Ecol. 65: 499–513.

NATIONAL ACADEMY OF SCIENCES. 1995. Understanding marinebiodiversity. National Academy Press.

NATIONAL RESEARCH COUNCIL. 1994. Priorities for coastal ecosys-tem science. National Academy Press.

NECKLES, H. A., R. L. WETZEL, AND R. J. ORTH. 1993. Relativeeffects of nutrient enrichment and grazing on epiphyte-mac-rophyte (Zostera marina) dynamics. Oecologia 93: 285–295.

NELSON, W. G. 1980. A comparative study of amphipods in seagrassfrom Florida to Nova Scotia. Bull. Mar. Sci. 30: 80–89.

NEUNDORFER, J. V., AND W. M. KEMP. 1993. Nitrogen versus phos-phorus enrichment of brackish waters: Responses of the sub-mersed plant Potamogeton perfoliatus and its associated algalcommunity. Mar. Ecol. Prog. Ser. 94: 71–82.

NIXON, S. W. 1995. Coastal marine eutrophication: A definition,social causes and future concerns. Ophelia 41: 199–219.

OFFICER, C. B. AND OTHERS. 1984. Chesapeake Bay anoxia: Origin,development, and significance. Science 223: 22–27.

ORTH, R. J., AND K. A. MOORE. 1983. Chesapeake Bay: An un-precedented decline in submerged aquatic vegetation. Science222: 51–53.

, AND J. VAN MONTFRANS. 1984. Faunal communities in sea-grass beds: A review of the influence of plant structure and

prey characteristics on predator-prey relationships. Estuaries 7:339–350.

, AND . 1987. Utilization of seagrass meadow andtidal marsh creek by blue crabs Callinectes sapidus. I. Seasonaland annual variations in abundance with emphasis on post-settlement juveniles. Mar. Ecol. Prog. Ser. 41: 283–394.

PAINE, R. T. 1966. Food web complexity and species diversity.Amer. Nat. 100: 65–75.

PARFIT, M. 1995. Diminishing returns. Nat. Geogr. 188: 2–36.PHILLIPART, C. J. M. 1995. Effect of periphyton grazing by Hydro-

bia ulvae on the growth of Zostera noltii on a tidal flat in theDutch Wadden Sea. Mar. Biol. 122: 431–437.

POLIS, G. A., AND D. S. STRONG. 1996. Food web complexity andcommunity dynamics. Am. Nat. 147: 813–846.

POWER, M. E. 1992. Top down and bottom up forces: Do plantshave primacy? Ecol. 73: 733–746.

ROBERTSON, A. I., AND K. H. MANN. 1982. Population dynamicsand life history adaptations of Littorina neglecta (Bean) in aneelgrass meadow (Zostera marina L.) in Nova Scotia. J. Exp.Mar. Biol. Ecol. 63: 151–171.

RUDLOE, A. 1985. Variation in the expression of lunar and tidalrhythms in the horseshoe crab, Limulus polyphemus. Bull. Mar.Sci. 36: 867–893.

SAFINA, C. 1995. The world’s imperiled fish. Sci. Amer. (Nov.): 46–53.

SHAPIRO, J., AND D. I. WRIGHT. 1984. Lake restoration by manip-ulation: Round Lake, Minnesota, the first two years. FreshwaterBiol. 14: 371–383.

SHORT, F. T., D. M. BURDICK, AND J. E. KALDY III. 1995. Mesocosmexperiments quantify the effects of eutrophication on eelgrass,Zostera marina. Limnol. Oceanogr. 40: 740–749.

SOGARD, S. M., G. V. N. POWELL, AND J. G. HOLMQUIST. 1987.Epibenthic fish communities on Florida Bay banks: Relationswith physical parameters and seagrass cover. Mar. Ecol. Prog.Ser. 40: 25–39.

STEWART, R. A., AND D. S. GORSLINE. 1962. Recent sedimentaryhistory of St. Joseph Bay, Florida. Sedimentology 1: 256–286.

STONER, A. W. 1982. The influence of benthic macrophytes on theforaging behavior of pinfish, Lagodon rhomboides. J. Exp.Mar. Biol. Ecol. 58: 271–284.

, AND R. J. LIVINGSTON. 1984. Ontogenetic patterns in dietand feeding morphology in sympatric sparid fishes from sea-grass meadows. Copeia 1: 174–187.

STRICKLAND, J. D. H., AND T. R. PARSONS. 1972. A practical hand-book for seawater analysis. Fish. Res. Bd. Can. Bull. 167. 310p.

STRONG, D. W. 1992. Are trophic cascades all wet? Differentiationand donor-control in speciose ecosystems. Eco. 73: 747–754.

TOMASKO, D. A., C. J. DAWES, AND M. O. HALL. 1996. The effectsof anthropogenic nutrient enrichment on turtlegrass (Thalassiatestudinum) in Sarasota Bay, Florida. Estuaries 19: 448–456.

, AND B. E. LAPOINTE. 1991. Productivity and biomass ofThalassia testudinum as related to water column nutrient avail-ability: Field observations and experimental studies. Mar. Ecol.Progr. Ser. 75: 9–17.

TWILLEY, R. R., W. M. KEMP, K. W. STAVER, J. C. STEVENSON,AND W. R. BOYNTON. 1985. Nutrient enrichment of estuarinesubmersed vascular plant communities. 1. Algal growth andeffects on production of plants and associated communities.Mar. Ecol. Progr. Ser. 23: 179–191.

VALENTINE, J. F., AND K. L. HECK, JR. 1991. The role of sea urchingrazing in regulating subtropical seagrass meadows: Evidenceform field manipulations in the northern Gulf of Mexico. J.Exp. Mar. Biol. Ecol. 154: 215–230.

, AND . 1993. Mussels in seagrass meadows: Theirinfluence on macroinvertebrate abundance and secondary pro-


duction in the northern Gulf of Mexico. Mar Ecol. Progr. Ser.96: 63–74.

VAN MONTFRANS, J., R. J. ORTH, AND S. A. VAY. 1982. Preliminarystudies of grazing by Bittium varium on eelgrass periphyton.Aquat. Bot. 14: 75–89.

, R. L. WETZEL, AND R. J. ORTH. 1984. Epiphyte-grazerrelationships in seagrass meadows: Consequences for seagrassgrowth and production. Estuaries 7: 298–309.

WERNER, E. E., AND J. F. GILLIAM. 1984. The ontogenetic nicheand species interactions in size-structured populations. Ann.Rev. Ecol. Syst. 15: 393–425.

WILLIAMS, A. H., L. D. COEN, AND M. S. STOELTING. 1990. Sea-sonal abundance, distribution, and habitat selection of juvenile

Callinectes sapidus (Rathbun) in the northern Gulf of Mexico.J. Exp. Mar. Biol. Ecol. 137: 165–183.

WILLIAMS, S. L., AND M. H. RUCKELSHAUS. 1993. Effects of nitro-gen availability and herbivory on eelgrass (Zostera marina)and epiphytes. Ecol. 74: 904–918.

ZARET, T. M. 1980. Predation in freshwater communities. YaleUniv. Press.

ZIMMERMAN, R., R. GIBSON, AND J. HARRINGTON. 1979. Herbivoryand detritivory among gammaridean amphipods from a Floridaseagrass community. Mar. Biol. 54: 41–47.

Received: 30 August 1999Accepted: 6 January 2000

Amended: 6 March 2000

Effects of nutrient enrichment and small predator density on seagrass ecosystems: An experimental assessment

Documents