marsh terracing as a wetland …courses.washington.edu/esrm479/marsh terracing.pdf327 wetlands, vol....

327

WETLANDS, Vol. 21, No. 3, September 2001, pp. 327–341q 2001, The Society of Wetland Scientists

MARSH TERRACING AS A WETLAND RESTORATION TOOL FOR CREATINGFISHERY HABITAT

Lawrence P. Rozas and Thomas J. MinelloNOAA/National Marine Fisheries Service

Southeast Fisheries Science Center4700 Avenue U

Galveston, Texas, USA 77551-5997E-mail: [email protected]

Abstract: Terracing is a relatively new wetland-restoration technique used to convert shallow subtidal bot-tom to marsh. This method uses existing bottom sediments to form terraces or ridges at marsh elevation. Aterrace field is constructed by arranging these ridges in some pattern that maximizes intertidal edge andminimizes fetch between ridges; the intertidal area is planted with marsh vegetation. We examined the habitatvalue of terracing for fishery species at Sabine National Wildlife Refuge, Louisiana (USA) in spring andfall 1999 by quantifying and comparing nekton densities in a 9-yr-old terrace field and nearby reference areausing a 1-m2 drop sampler. Decapod crustaceans were more abundant than fishes, composing 62% and 95%of all organisms we collected in spring and fall, respectively. White shrimp Litopenaeus setiferus, dagger-blade grass shrimp Palaemonetes pugio, blue crab Callinectes sapidus, and brown shrimp Farfantepenaeusaztecus accounted for 94% of all crustaceans, whereas 60% of all fishes were gulf menhaden Brevoortiapatronus. Mean densities of white shrimp (fall), daggerblade grass shrimp, blue crab, and brown shrimp(spring) were significantly greater in terrace marsh than on non-vegetated bottom in the reference pond.Densities of most nekton on non-vegetated bottom were similar in the terrace field and the reference pond,but gulf menhaden and white shrimp had higher densities at terrace pond sites and brown shrimp (spring)were more abundant at reference pond sites. The pattern for biomass was similar to that for density in thatthe mean biomass of most species was significantly greater at terrace marsh sites than reference pond sitesand similar at terrace and reference pond sites. Terrace marsh, however, was not functionally equivalent tonatural marsh, as mean densities of daggerblade grass shrimp (fall), brown shrimp (spring), and blue craband mean biomass of white shrimp (fall), striped mullet Mugil cephalus (spring), and spotted seatrout Cy-noscion nebulosus (fall) were greater at reference marsh sites than terrace marsh sites. Using these densityand biomass patterns and the percentage of marsh and pond area in the terrace field, we concluded thatterrace fields support higher standing crops of most fishery species compared with shallow marsh ponds ofsimilar size. Future restoration projects could include design changes to increase the proportion of marsh ina terrace field and enhance the habitat value of marsh terraces for fishery species.

Key Words: fishery species, Gulf of Mexico, habitat comparisons, habitat selection, nursery areas, penaeidshrimps, tidal marsh, restoration

INTRODUCTION

Estuarine marshes, and in particular the vegetationat the marsh-water interface (marsh edge), harbordense concentrations of young fishery species (Zim-merman and Minello 1984, Baltz et al. 1993, Cicchetti1998, Howe et al. 1999, Rozas and Zimmerman 2000).As nursery areas, these wetlands support valuablecoastal fisheries along the Gulf of Mexico and South-east Atlantic coasts of the U.S. (Kneib 1997, Zimmer-man et al. 2000). Over the last 50 years, the total areaof coastal marsh habitat has decreased substantially aswetlands were filled for development or converted toopen water through submergence (Mitsch and Gosse-link 1986).

In response to this large-scale loss of coastal wet-lands, a variety of techniques have been used to restorethese estuarine habitats. Structural marsh management,or the use of impoundments with water-control struc-tures, has been used as a wetland restoration techniquethat may be effective in restoring wetland vegetationin certain oligohaline areas (see Boyer 1997 for anevaluation of effectiveness), but evidence is strong thatthis approach has negative effects on habitat use bytransient fishery species (McGovern and Wenner 1990,Rogers et al. 1994, Rozas and Minello 1999). In directcontrast, wetland restoration through re-establishmentof tidal flow to diked marsh appears to positively affecthabitat use by fishery species (Gilmore et al. 1982, Reyet al. 1990, Chamberlain and Barnhart 1993, Peck et

328 WETLANDS, Volume 21, No. 3, 2001

Figure 1. Aerial photograph of a terrace field at Sabine National Wildlife Refuge showing terraces laid out in a checkerboardpattern. Corners are open to allow movement of water and organisms through the terrace field. (Photograph taken January 16,2001 by Ron Boustany and Tommy Michot.)

al. 1994). Perhaps the most common restoration tech-nique, however, simply involves planting marsh veg-etation on dredged material or on scraped-down up-lands. In the southeastern United States, this approachappears to restore some habitat functions for estuarinenekton, but there is also evidence that these createdmarshes are not functionally equivalent to naturalmarshes (Moy and Levin 1991, Minello and Zimmer-man 1992, Meyer et al. 1993, Minello and Webb 1997,Minello 2000, Streever 2000). One important reasonfor this deficiency is that these created marshes (es-pecially those constructed on fine-grained dredged ma-terial) have relatively few creeks and ponds, and con-sequently little edge habitat (Minello et al. 1994, De-laney et al. 2000, Streever 2000).

Marsh terracing is a relatively new wetland-resto-ration technique that creates a substantial amount ofmarsh edge (Underwood et al. 1991). Terraces areridges or levees of sediment at marsh elevation thatare constructed by excavating subtidal bottom sedi-ments on site. The borrow areas created by terraceconstruction trap and eventually become filled withsediment. Terrace fields are developed by arranging aseries of these ridges in some pattern that maximizesintertidal edge and minimizes fetch between ridges.One arrangement is a checkerboard pattern with opencorners (Figure 1). Following construction, the inter-tidal levees are planted with marsh vegetation; and thesubtidal areas between terrace levees, may be plantedwith seagrasses or other species of submerged aquatic

vegetation (SAV). Marsh terracing has been promotedas a means of enhancing deposition and retention ofsuspended sediments, reducing turbidity, increasingmarsh-edge habitat, increasing overall primary andsecondary productivity, and maximizing access formarine organisms (Underwood et al. 1991, LDNR1993). Currently, marsh terracing appears to be re-stricted to restoration efforts along the northern Gulfof Mexico coast. However, the technique could be ap-plied in other areas with shallow water and suitablesubstrates. For example, land-reclamation methodsused for decades in Europe are quite similar to thetechniques used in marsh terracing (Kamps 1962).

The first terracing project in the U.S. was construct-ed at the Sabine National Wildlife Refuge (SabineNWR), Cameron Parish (county), Louisiana in 1990(Underwood et al. 1991, LDNR 1993). Spartina alter-niflora Loisel. was successfully established along theterrace levees in October 1990, and this species dom-inated well-established stands of marsh vegetation onthe terraces when we initiated our study in 1999. Aninitial attempt to establish SAV was apparently unsuc-cessful; three species of SAV (Ruppia maritima Lin-naeus, Halodule wrightii Ascherson, and Thalassiatestudinum Konig) were planted in experimental plotswithin the newly-constructed terrace ponds in June1991, but few plants had survived by August 1991(LDNR 1993).

A major goal of marsh terracing is to provide habitatfor fishery species, but the habitat value of terrace

Rozas & Minello, HABITAT VALUE OF MARSH TERRACING 329

fields has not been assessed. Specific objectives of ourstudy were to document the use of marsh terraces bynekton (dominant species of fishes and decapod crus-taceans) and to compare densities and biomasses offishery species (as a measure of habitat quality) in aterraced area and a nearby reference area at SabineNWR. Our use of animal density and biomass to mea-sure habitat quality is predicated on the assumptionthat animal distributions reflect the capacity of an areato provide essential habitat functions for species. Theresults from our research can be used to predict chang-es in habitat value that may be anticipated in areaswhere new terracing projects are now being planned.

MATERIALS AND METHODS

Study Area

The study area is in Southwest Louisiana within theSabine NWR and located on Calcasieu Lake (Figure2). Salinities at the study area are 10–15‰ in springand 15–20‰ in fall during years of average rainfall(Orlando et al. 1993). Tides in the estuary are predom-inantly diurnal and have a mean daily range of ,0.3m (U.S. Department of Commerce 1998). Salt marshdominated by Spartina alterniflora occupies much ofthe low intertidal zone within the study area.

Terrace fields were built in 1990 within two of threeshallow marsh ponds located on the north shore ofWest Cove, an arm of Calcasieu Lake (Figure 2).These terrace fields were constructed in a checker-board pattern with cells having 61-m-long sides (0.37-ha pond area). Intertidal terrace ridges within cellswere approximately 55 m long. We selected the pondwithout terraces as a reference area and the adjacentterraced pond as the two areas for comparison in ourstudy design (Figure 2). We identified four treatmentsfor sampling: two treatments were intertidal and veg-etated with Spartina alterniflora (terrace marsh andreference marsh) and two were subtidal open-water,mud bottom (terrace pond and reference pond). Mostof the shallow, open-water areas did not support SAV,but a few areas contained Ruppia maritima.

In each of two seasons (spring and fall), we col-lected a total of 60 samples from randomly selectedsites at the terrace and reference areas; the sites wereselected using random numbers and a grid placed overan aerial photograph. We took ten Spartina marshedge samples each on terrace levees (terrace marsh)and along the shoreline of the reference pond (refer-ence marsh). In addition, 20 samples each were takenin terrace ponds and the reference pond. We collectedtwice as many open-water samples as marsh samplesto ensure that at least some sites would contain SAV.Further, more samples are required to accurately rep-

resent non-vegetated areas because these sites containrelatively few organisms and variability among sites ishigh. We collected all samples in the day at high tideduring periods of tropical tides May 4–5 and Septem-ber 7–8 and 20, 1999.

Nekton Sampling

Fishes and decapod crustaceans (nekton) were quan-titatively sampled with a drop sampler using the pro-cedure described by Zimmerman et al. (1984). Wechose a drop sampler for this study because it is ef-fective in dense emergent vegetation, and the catchefficiency of this enclosure device does not appear tovary substantially with habitat characteristics typical ofshallow estuarine areas (Rozas and Minello 1997). Thesampler was a 1.14-m-diameter cylinder that wedropped from a boom attached to a shallow-draft boat.Two persons positioned the cylinder over a sample siteby slowly pushing from the boat’s stern. When re-leased from the boom, the cylinder rapidly entrappedorganisms within a 1.0-m2 sample area.

After the cylinder was dropped, we measured watertemperature, dissolved oxygen, salinity, and turbidityusing the methods described by Minello and Zimmer-man (1992). We determined water depth at each sam-ple site by averaging five depth measurements takenwithin the sampler. We also measured the distancefrom the sample area to the nearest marsh-water inter-face. At marsh sites, plant stems were clipped atground level, counted (dead and alive combined), andremoved from the cylinder. If SAV was present at ter-race or reference pond sites, we estimated coveragewithin the sampler (0–100%) and identified the speciesof plants present.

After measuring the environmental variables, wecaptured nekton trapped in the drop sampler by usingdip nets and filtering the water pumped out of the en-closure through a 1-mm-mesh net. When the samplerwas completely drained, any animals remaining on thebottom were removed by hand. Samples were pre-served in formalin and returned to the laboratory forprocessing.

In the laboratory, the samples were sorted, and an-imals were identified to lowest feasible taxon. We usedthe nomenclature of Perez-Farfante and Kensley(1997) for penaeid shrimps and identified species us-ing the protocol described in Rozas and Minello(1998). Thirty-one specimens of Farfantepenaeuscould not be reliably identified either because of theirsize (total length 13–18 mm) or because they weredamaged; these shrimps were assigned as brownshrimp F. aztecus (Ives) or pink shrimp F. duorarum(Burkenroad) based on the proportion of identifiedspecies in each sample. An additional 79 damaged


Table 1. Analysis of Variance (ANOVA) table for comparing the four treatment areas. Model includes the test for the main effect ofTreatment and the a priori contrasts that compare specific treatment effects. The dependent variable used in the example presented hereis total macrofauna (sum of total fishes and crustaceans).

Source dfSum ofSquares

MeanSquare F Value P Value

May 1999TREATMENT

CONTRASTSTerrace Marsh vs Reference PondTerrace Pond vs Reference PondTerrace Marsh vs Reference Marsh

RESIDUAL ERROR

3

111

56

58.451

29.8680.0520.003

64.142

19.484

29.8680.0520.0031.145

17.011

26.0760.0460.003

0.0001

0.00010.83160.9592

September 1999TREATMENT

CONTRASTSTerrace Marsh vs Reference PondTerrace Pond vs Reference PondTerrace Marsh vs Reference Marsh

RESIDUAL ERROR

3

111

56

55.395

25.8442.3521.204

18.098

18.465

25.8442.3521.2040.323

57.136

79.9697.2773.725

0.0001

0.00010.00920.0587

←

Figure 2. Map of the study area and its location on the north shore of West Cove in Southwest Louisiana. Study area islocated within the Calcasieu Lake estuary at approximately 29o 53’ N latitude and 93o 23’ W longitude.

shrimp could only be identified as penaeids and wereassigned to species (white shrimp Litopenaeus setifer-us (Linnaeus), brown shrimp, or pink shrimp) usingthis same procedure. Grass shrimp (146) that could notbe identified to species were similarly assigned to oneof two species (daggerblade grass shrimp Palaemo-netes pugio Holthuis, or brackish grass shrimp P. in-termedius Holthuis) identified from our samples. An-imals that could not be reliably identified, either dueto their size or because they were damaged, were notused in size analyses. Total length of fishes andshrimps and carapace width of crabs were measuredto the nearest mm. The biomass for each species wasdetermined by pooling individuals in a sample andmeasuring wet weight to the nearest 0.1g.

As a wetland restoration technique, terrace fields arebuilt to replace shallow open water in marsh ponds.Therefore, we also used our density data to comparestanding crops (in numbers of organisms and biomassfor selected taxa) between a 1-ha terrace field and areference pond of the same size. We used recent aerialphotography to calculate that 21% of the terraced areawas occupied by marsh and 79% by pond. We rec-ognize that a portion of each terrace ridge may havebeen exploited very little by aquatic organisms becausethe middle of the ridge was relatively high in eleva-tion. For this exercise, however, we assumed that theentire terrace ridge was equally used by fishery spe-cies. We estimated standing crops of animals within

the two 1-ha areas for May and September using av-erage animal densities and biomass determined fromour nekton samples and the total area of each habitattype within the terrace field and reference pond. Wealso used these data to simulate the effect of increasingthe proportion of marsh in a terrace field on brownshrimp standing crops (total numbers) in May.

Data Analyses

We used 1-way Analysis of Variance (ANOVA)followed by a priori contrasts to examine differencesin densities and biomass of selected organisms and en-vironmental characteristics (mean dissolved oxygen,salinity, water temperature, turbidity, water depth, anddistance to edge) among treatments (terrace marsh, ref-erence marsh, terrace pond, and reference pond) (Table1). We compared the following treatments with apriori contrasts (Table 1): Terrace Marsh vs ReferencePond, Terrace Pond vs Reference Pond, and TerraceMarsh vs Reference Marsh. These first two contrastswere selected because terrace marsh and terrace pondare habitat types created from the reference pond hab-itat type when terraces are constructed. The terracemarsh-reference marsh contrast was used to determinewhether created marsh in the terrace field was com-parable to natural marsh in supporting nekton.

In the ANOVA procedure, we analyzed the data col-lected during each season separately, because several


species were only abundant enough to include in thestatistical analysis in one season. We considered alphalevels of 0.05 to be significant in all results, but wealso assessed significance after adjusting alpha levelsfor the Treatment effect using the sequential Bonfer-roni method described by Rice (1989), which buffersagainst error introduced by making multiple compari-sons with the same sample set (i.e., testing a hypoth-esis for several species or variables). Densities andbiomass of animals were positively related to the stan-dard deviation; therefore, we performed a ln (x 1 1)transformation of the original values prior to analyses.Other variables were not transformed. All tabular andgraphical data presented in this paper are untrans-formed means. We conducted statistical analyses usingSuperANOVA (Version 5 Ed., Abacus Concepts, Inc.,Berkeley, California, 1989).

RESULTS

We collected a total of 1,344 animals (21 fish and9 decapod crustacean species) with a biomass of 2.21kg wet weight in May and 3,575 animals (18 fish and12 decapod crustacean species) and 0.8 kg wet weightin September (Tables 2 and 3). Fishes accounted for58% of the total biomass in May, whereas decapodcrustaceans composed most (83%) of the biomass inSeptember. Crustaceans outnumbered fishes in bothsampling periods, accounting for 62% and 95% of allorganisms taken in May and September, respectively.Species of decapod crustaceans in our samples wererelatively consistent between sampling periods (Table2). Daggerblade grass shrimp, brown shrimp, blue crabCallinectes sapidus Rathbun, brackish grass shrimp,and pink shrimp accounted for 98% of the crustaceanstaken in May. These same species, along with whiteshrimp, composed 99% of the crustaceans collected inSeptember. Other taxa infrequently taken in our sam-ples included the bigclaw snapping shrimp Alpheusheterochaelis Say, the lesser blue crab Callinectes sim-ilis Williams, the glabrous broadface crab Euryplaxnitida Stimpson, the xanthid crabs Rhithropanopeusharrisii (Gould) and Dyspanopeus texana (Stimpson),and the fiddler crabs Uca longisignalis Salmon andAtsaides and Uca rapax (Smith).

Abundant fishes differed between May and Septem-ber (Table 2). Most (86% of total) fishes taken in Maywere gulf menhaden Brevoortia patronus Goode andspot Leiostomus xanthurus Lacepede, whereas nakedgoby Gobiosoma bosc (Lacepede), code goby Gobio-soma robustum Ginsburg, and unidentified gobies(Family Gobiidae) were most abundant in September.Other less abundant but commonly collected speciesincluded Anchoa mitchilli (Valenciennes), Cynoscionnebulosus (Cuvier), Cyprinodon variegatus Lacepede,

Fundulus grandis Baird and Girard, Gobiesox stru-mosus Cope, Gobionellus boleosoma (Jordan and Gil-bert), Menidia beryllina (Cope), Micropogonias un-dulatus (Linnaeus), and Mugil cephalus Linnaeus. Inaddition, we infrequently collected (,5 individuals)Adinia xenica (Jordan and Gilbert), Citharichthys spi-lopterus Gunther, Cynoscion arenarius Ginsburg,Etropus crossotus Jordan and Gilbert, Evorthodus lyr-icus (Girard), Fundulus pulvereus (Evermann), Fun-dulus similis (Baird and Girard), Gobionellus oceani-cus (Pallas), Harengula jaguana Poey, Lagodon rhom-boides (Linnaeus), Lucania parva (Baird and Girard),Mugil curema Valenciennes, Myrophis punctatus Lut-ken, Paralichthys lethostigma Jordan and Gilbert, Poe-cilia latipinna (Lesueur), Sciaenops ocellatus (Linnae-us), Stellifer lanceolatus (Holbrook), and Symphurusplagiusa (Linnaeus) in the study area.

Mean densities of decapod crustaceans were greatestin intertidal marsh, and most species were significantlymore abundant at terrace marsh sites than referencepond sites (Table 2). Exceptions were brown shrimp(in September only) and brackish grass shrimp; den-sities at terrace marsh sites and reference pond sitesfor these species were not significantly different. Crus-tacean densities were generally low over all non-veg-etated bottom, and mean densities of most specieswere not significantly different between terrace pondand reference pond sites. Exceptions to this generalitywere brown shrimp (in May), which were more abun-dant at reference pond sites, and white shrimp (in Sep-tember), which were more abundant in terrace ponds.Four crustacean species had mean densities in the ter-race marsh that were only 11–54% of their densitiesin the reference marsh (brown shrimp519% and 54%,blue crab532% and 44%, brackish grass shrimp511%and 18%, and daggerblade grass shrimp in Septem-ber554%). Densities of daggerblade grass shrimp inMay and white shrimp in September, however, werenot significantly different between marsh treatments.Fishes did not follow such a consistent spatial pattern.Gulf menhaden was most abundant over non-vegetatedbottom, and mean densities of this species were greaterat terrace pond than reference pond sites (Table 2).Mean densities of naked goby and code goby weregreatest in the reference marsh; however, most un-identified gobies were taken in the reference pond (Ta-ble 2).

The pattern for biomass was similar to the densitypattern for most numerically-dominant species (Table3). Biomass and density patterns differed, however, forbrown shrimp, blue crab, and white shrimp. Mean bio-mass of brown shrimp (May) was greater at terracemarsh sites than reference pond sites and did not differsignificantly between terrace pond sites and referencepond sites (Table 3). Mean blue crab biomass at ref-


Tabl

e2.

Com

pari

son

ofde

nsiti

es(m

ean

m2

26

1SE

)of

the

mos

tab

unda

ntde

capo

dcr

usta

cean

san

dfis

hes

colle

cted

amon

gtr

eatm

ents

(ter

race

mar

sh,

refe

renc

em

arsh

,te

rrac

epo

nd,

and

refe

renc

epo

nd)

inM

ayan

dSe

ptem

ber

1999

.E

ach

mea

nis

estim

ated

from

10(m

arsh

)or

20(p

ond)

sam

ples

.T

heto

tal

num

ber

ofsp

ecie

sco

llect

edin

each

taxo

nom

icca

tego

ryal

sois

give

n.R

esul

ts(p

valu

es)

are

give

nfo

rA

NO

VA

anal

yses

we

used

toco

mpa

rem

ean

dens

ities

amon

gth

efo

urtr

eatm

ent

area

san

da

prio

rico

ntra

sts

test

ing

for

sign

ifica

ntdi

ffer

ence

sbe

twee

n:1

5te

rrac

em

arsh

and

refe

renc

epo

nd;

25

terr

ace

pond

and

refe

renc

epo

nd;

and

35

terr

ace

mar

shan

dre

fere

nce

mar

sh.

The

AN

OV

Am

odel

used

todo

thes

ean

alys

esis

show

nin

Tabl

e1.

An

*in

dica

tes

that

the

prob

abili

tyva

lue

was

sign

ifica

ntfo

rth

eov

eral

lA

NO

VA

afte

ral

pha

was

adju

sted

asde

scri

bed

byR

ice

(198

9).

Spec

ies

Terr

ace

Mar

sh

Mea

nSE

Ref

eren

ceM

arsh

Mea

nSE

Terr

ace

Pond

Mea

nSE

Ref

eren

cePo

nd

Mea

nSE

Tota

l%

Tota

lA

NO

VA

pV

alue

Con

tras

tp

Val

ues

(1)

TM

arsh

vsR

Pond

(2)

TPo

ndvs

RPo

nd

(3)

TM

arsh

vsR

Mar

sh

May

1999

Cru

stac

eans

(Tot

al5

9sp

ecie

s)D

agge

rbla

degr

ass

shri

mp

Bro

wn

shri

mp

Blu

ecr

abB

rack

ish

gras

ssh

rim

p

26.6 4.6

3.1

1.3

(6.5

2)(0

.40)

(0.5

3)(0

.91)

16.0 8.5

7.0

7.4

(2.4

7)(1

.15)

(1.5

9)(3

.54)

0.0

0.6

0.2

0.0

(0.0

0)(0

.17)

(0.0

8)(0

.00)

0.2

1.6

0.5

0.3

(0.1

1)(0

.50)

(0.2

0)(0

.30)

429

174

114 93

51.3

20.8

13.6

11.1

0.00

01*

0.00

01*

0.00

01*

0.00

06*

0.00

010.

0001

0.00

010.

2241

0.57

350.

0096

0.27

900.

6691

0.29

470.

0321

0.03

240.

0275

Pink

shri

mp

Tota

lC

rust

acea

nsFi

shes

(Tot

al5

21sp

ecie

s)G

ulf

men

hade

nSp

otTo

tal

Fish

es

0.8

37.1 2.6

0.8

5.5

(0.5

9)(6

.81)

(1.8

7)(0

.49)

(2.3

2)

0.3

39.7 0.0

0.2

2.1

(0.3

0)(4

.94)

(0.0

0)(0

.13)

(0.4

3)

0.0

0.8

15.3 0.2

15.9

(0.0

0)(0

.19)

(8.0

0)(0

.08)

(8.0

5)

0.1

2.7

4.4

0.3

5.7

(0.0

5)(1

.01)

(3.9

9)(0

.16)

(3.9

8)

13 837

418 19 507

1.6

82.4 3.7

0.00

01*

0.03

68

0.57

80

0.00

01

0.53

78

0.00

66

0.02

21

0.50

69

0.23

81

Sept

embe

r19

99C

rust

acea

ns(T

otal

512

spec

ies)

Whi

tesh

rim

pD

agge

rbla

degr

ass

shri

mp

Blu

ecr

ab

62.1

38.0 4.7

(11.

92)

(7.8

4)(0

.65)

64.2

70.8

14.6

(8.8

4)(1

3.02

)(1

.56)

19.6 0.4

2.6

(2.6

7)(0

.22)

(0.4

4)

7.8

0.2

2.0

(1.1

1)(0

.08)

(0.3

5)

1810

1098 285

53.5

32.4 8.4

0.00

01*

0.00

01*

0.00

01*

0.00

010.

0001

0.00

17

0.00

060.

6551

0.33

08

0.64

410.

0023

0.00

01B

rack

ish

gras

ssh

rim

pB

row

nsh

rim

pPi

nksh

rim

pTo

tal

Cru

stac

eans

Fish

es(T

otal

18sp

ecie

s)

0.8

0.6

0.0

107.

2

(0.7

0)(0

.40)

(0.0

0)(1

8.08

)

7.0

3.1

1.4

162.

4

(3.7

7)(0

.80)

(1.4

0)(1

9.22

)

0.6

0.2

0.2

23.5

(0.6

0)(0

.16)

(0.0

8)(3

.15)

0.0

0.6

0.2

10.9

(0.0

0)(0

.20)

(0.2

0)(1

.45)

90 52 2233

84

2.7

1.5

0.00

24*

0.00

01*

0.00

01*

0.32

950.

9236

0.00

01

0.57

950.

1721

0.00

13

0.01

510.

0001

0.06

99

Uni

dent

ified

goby

Nak

edgo

byC

ode

goby

Gul

fki

llifis

hTo

tal

Fish

es

0.1

0.0

0.0

1.1

3.1

(0.1

0)(0

.00)

(0.0

0)(0

.50)

(0.9

7)

0.5

0.9

1.0

0.2

4.9

(0.2

7)(0

.46)

(0.6

7)(0

.20)

(0.9

4)

0.2

0.3

0.0

0.0

1.8

(0.1

2)(0

.30)

(0.0

0)(0

.00)

(0.8

1)

1.5

0.6

0.7

0.0

3.8

(0.5

6)(0

.36)

(0.3

0)(0

.00)

(0.6

8)

39 26 23 13 191

20.4

13.6

12.0 6.8

0.00

82*

0.16

950.

0657

0.00

07*

0.00

58

0.38

53

0.00

26

0.00

09

0.32

19

0.10

51


Tabl

e3.

Com

pari

son

ofbi

omas

ses

(mea

n6

1SE

)in

gram

sof

dom

inan

t(c

ontr

ibut

ing

mos

tbi

omas

s)de

capo

dcr

usta

cean

san

dfis

hes

colle

cted

amon

gtr

eatm

ents

(ter

race

mar

sh,

refe

renc

em

arsh

,te

rrac

epo

nd,

and

refe

renc

epo

nd)

inM

ayan

dSe

ptem

ber

1999

.E

ach

mea

nis

estim

ated

from

10(m

arsh

)or

20(p

ond)

sam

ples

.R

esul

ts(p

valu

es)

are

give

nfo

rA

NO

VA

anal

yses

we

used

toco

mpa

rem

ean

biom

asse

sam

ong

the

four

trea

tmen

tar

eas

and

apr

iori

cont

rast

ste

stin

gfo

rsi

gnifi

cant

diff

eren

ces

betw

een:

15

terr

ace

mar

shan

dre

fere

nce

pond

;2

5te

rrac

epo

ndan

dre

fere

nce

pond

;an

d3

5te

rrac

em

arsh

and

refe

renc

em

arsh

.T

heA

NO

VA

mod

elus

edto

doth

ese

anal

yses

issh

own

inTa

ble

1.A

n*

indi

cate

sth

atth

epr

obab

ility

valu

ew

assi

gnifi

cant

for

the

over

all

AN

OV

Aaf

ter

alph

aw

asad

just

edas

desc

ribe

dby

Ric

e(1

989)

.

Spec

ies

Terr

ace

Mar

sh

Mea

nSE

Ref

eren

ceM

arsh

Mea

nSE

Terr

ace

Pond

Mea

nSE

Ref

eren

cePo

nd

Mea

nSE

Tota

l%

Tota

lA

NO

VA

pV

alue

Con

tras

tp

Val

ue

(1)

TM

arsh

vsR

Pond

(2)

TPo

ndvs

RPo

nd

(3)

TM

arsh

vsR

Mar

sh

May

1999

Cru

stac

eans

Blu

ecr

abB

row

nsh

rim

pD

agge

rbla

degr

ass

shri

mp

39.1 2.7

7.4

(26.

73)

(0.8

6)(2

.01)

3.8

9.7

5.4

(0.8

9)(1

.83)

(0.9

9)

8.3

0.7

0.0

(8.2

5)(0

.25)

(0.0

0)

0.1

1.1

0.0

(0.0

3)(0

.32)

(0.0

1)

596.

515

9.8

128.

1

64.8

17.3

13.9

0.00

01*

0.00

01*

0.00

01*

0.00

010.

0235

0.00

01

0.48

080.

2923

0.93

08

0.32

200.

0001

0.73

08B

rack

ish

gras

ssh

rim

pPi

nksh

rim

pTo

tal

Cru

stac

eans

Fish

es

0.2

0.0

50.8

(0.1

4)(0

.04)

(26.

67)

0.8

0.3

20.7

(0.4

1)(0

.29)

(2.2

8)

0.0

0.0

9.1

(0.0

0)(0

.00)

(8.3

0)

0.0

0.0

1.2

(0.0

3)(0

.00)

(0.3

3)

10.5 3.3

921.

0

1.1

0.00

10*

0.00

01*

0.21

60

0.00

01

0.80

21

0.87

78

0.03

24

0.84

05

Stri

ped

mul

let

Gul

fm

enha

den

Spot

Tota

lFi

shes

4.7

0.3

3.3

10.2

(4.7

1)(0

.18)

(1.8

3)(6

.55)

74.8 0.0

1.9

80.8

(39.

04)

(0.0

0)(1

.41)

(39.

09)

2.1

11.1 0.8

14.4

(2.1

4)(5

.92)

(0.4

3)(6

.05)

0.0

3.0

0.7

4.6

(0.0

0)(2

.92)

(0.4

8)(2

.93)

837.

428

6.2

81.5

1288

.9

65.0

22.2 6.3

0.00

11*

0.01

190.

2756

0.00

51*

0.43

130.

7838

0.34

61

0.63

770.

0109

0.16

78

0.00

600.

7046

0.01

82Se

ptem

ber

1999

Cru

stac

eans

Whi

tesh

rim

pB

lue

crab

Dag

gerb

lade

gras

ssh

rim

p

8.9

7.5

2.4

(2.0

5)(3

.62)

(0.6

4)

25.0 3.4

6.7

(5.1

3)(0

.65)

(2.0

4)

2.4

0.3

0.0

(0.3

6)(0

.09)

(0.0

3)

0.8

0.1

0.0

(0.2

4)(0

.03)

(0.0

1)

403.

511

5.9

92.3

62.1

17.8

14.2

0.00

01*

0.00

01*

0.00

01*

0.00

010.

0001

0.00

01

0.00

170.

6123

0.86

70

0.00

020.

8453

0.00

11B

rack

ish

gras

ssh

rim

pB

row

nsh

rim

pPi

nksh

rim

pTo

tal

Cru

stac

eans

Fish

es

0.0

0.1

0.0

18.8

(0.0

3)(0

.04)

(0.0

0)(3

.58)

1.5

1.0

0.2

37.5

(1.0

6)(0

.29)

(0.2

0)(7

.20)

0.0

0.0

0.0

2.7

(0.0

5)(0

.02)

(0.0

1)(0

.39)

0.0

0.2

0.0

1.2

(0.0

0)(0

.16)

(0.0

3)(0

.28)

16.5

15.6 2.8

649.

7

2.5

2.4

0.00

35*

0.00

01*

0.00

01*

0.77

120.

5513

0.00

01

0.75

820.

2451

0.00

22

0.00

540.

0001

0.00

68

Whi

tem

ulle

tSt

ripe

dm

ulle

tSp

otte

dse

atro

utG

ulf

killi

fish

Uni

dent

ified

goby

Tota

lFi

shes

0.0

2.6

0.0

1.0

0.3

5.1

(0.0

0)(1

.97)

(0.0

0)(0

.52)

(0.3

2)(2

.35)

3.1

0.0

1.8

0.2

0.0

6.8

(3.1

1)(0

.00)

(1.3

8)(0

.23)

(0.0

2)(3

.22)

0.0

0.0

0.0

0.0

0.0

0.2

(0.0

0)(0

.00)

(0.0

0)(0

.00)

(0.0

0)(0

.11)

0.0

0.0

0.0

0.0

0.0

0.3

(0.0

0)(0

.00)

(0.0

5)(0

.00)

(0.0

2)(0

.08)

31.1

26.0

18.6

12.3 4.3

129.

0

24.1

20.2

14.4 9.5

3.3

0.17

160.

0165

0.00

780.

0021

*0.

2690

0.00

01*

0.00

420.

8157

0.00

05

0.00

07

1.00

000.

8006

1.00

00

0.79

64

0.01

230.

0056

0.02

15

0.62

58


Figure 3. Comparison of hypothetical 1-ha terrace field and reference pond in May and September 1999 using estimatedstanding crop (total number of organisms) of abundant taxa. Animal densities used in the calculations are mean densitiesdetermined from sample collections in each habitat type (terrace marsh and terrace pond, reference pond). Design of the terracefield in this comparison is based on the marsh terraces in our study area where marsh and ponds occupy 21% and 79% of thetotal area, respectively.

erence marsh sites and terrace marsh sites was not sig-nificantly different, and mean biomass of white shrimp(September) was significantly greater at referencemarsh sites than terrace marsh sites (Table 3). Thereference marsh contributed most of the biomass ofthree fish species (striped mullet, white mullet, andspotted seatrout) that accounted for a substantial per-centage of the total biomass, even though they werenot abundant in our samples (Table 3). The mean bio-mass of striped mullet was significantly greater at ref-erence marsh sites than terrace marsh sites in May,although this species was collected exclusively at ter-race marsh sites in September (Table 3). Spotted sea-trout was taken almost exclusively in the referencemarsh, and white mullet was collected only in the ref-erence marsh (Table 3).

In comparisons between 1-ha areas of marsh terraceand reference pond, standing crops of most animalswere greater within the terrace field (Figures 3 and 4).Based on our density data for May, gulf menhaden,daggerblade grass shrimp, blue crab, and pink shrimpwere 2.9, 37.2, 1.5, and 3.4 times more abundant, re-spectively, in the terrace field than the reference pond.In September, white shrimp, daggerblade grass shrimp,and blue crab were 3.7, 55.0, and 1.5 times more abun-dant, respectively, in the terrace field. In contrast,standing crops (total numbers) of brown shrimp in theterrace field in May were 88% of those in the referencepond (Figure 3). Naked goby also was less abundantin the terrace field than in the reference pond. Com-parisons of standing crops using biomass data fol-lowed a similar pattern (Figure 4). The terrace field


Figure 4. Comparison of hypothetical 1-ha terrace field and reference pond in May and September 1999 using estimatedstanding crop (total biomass of organisms, kg) of selected species. Animal biomasses used in the calculations are meansdetermined from sample collections in each habitat type (terrace marsh and terrace pond, reference pond). Design of the terracefield in this comparison is based on the marsh terraces in our study area where marsh and ponds occupy 21% and 79% of thetotal area, respectively.

supported much greater biomasses than the referencepond for most species. Exceptions were brown shrimp(mean biomass in terrace field and reference pond wassimilar in May) and spotted seatrout (not collected ineither terrace habitat type in September).

Our results indicate that redesigning terrace fields toincrease the proportion of emergent marsh can enhancestanding crops of species that select for marsh vege-tation. Our estimate of brown shrimp standing crop inMay was slightly lower in 1 ha of Sabine terrace fieldthan in a marsh pond of equal size (Figures 3 and 5).When we simulated the effect of changing the area ofmarsh in a terrace field, we found that increasing themarsh area from 21% to 26% increased the standingcrop of brown shrimp to the level of the marsh pond(Figure 5).

Few of the species we tested differed in size amongtreatments. Within marsh treatments for September,white shrimp were significantly larger in the referencemarsh, whereas blue crab were larger in the terracemarsh (Table 4). Most species were taken too infre-quently in terrace ponds and the reference pond to in-clude these treatments in the size analysis. However,brown shrimp were abundant enough in the referencepond in May, and white shrimp and blue crab werecollected frequently enough in both terrace ponds andthe reference pond in September for analysis. In May,brown shrimp did not differ significantly in size amongtreatment areas (ANOVA, d.f.52,30, F51.677,p50.2039). In September, the mean size (20.1 mm) ofwhite shrimp at reference pond sites was not signifi-cantly different from those at either terrace marsh


Figure 5. An illustration of how increasing the proportionof marsh in a terrace field affects standing crop (total num-bers) of brown shrimp in spring. Animal densities used inthe calculations are mean brown shrimp densities determinedfrom May 1999 samples collected in each habitat type (ter-race marsh, terrace pond, and reference pond). When 21%of the terrace field is marsh (as in our study area), brownshrimp standing crop is less in the terrace field than in amarsh pond of equal size. Standing crop in the terrace fieldequals that in a marsh pond when 26% of the terrace fieldis marsh and exceeds the pond standing crop when . 26%of the terrace field is marsh.

Table 4. Comparison of sizes (mean 6 1 SE) in mm of numerically dominant animals collected at marsh sites in May and September1999. Each mean (total length of shrimps or carapace width of crabs) for a marsh treatment was estimated from the mean sizes of nsamples that contained that species. Degrees of freedom (df), F- and P-values listed are from an ANOVA comparing mean size in thetwo marsh treatments. An * indicates that the probability value was significant after alpha was adjusted as described by Rice (1989).

Species

Terrace Marsh

Mean SE n

Reference Marsh

Mean SE n df F P

MayBrown shrimpBlue crabDaggerblade grass shrimp

35.929.326.0

(4.64)(8.79)(1.53)

109

10

46.117.429.7

(2.90)(1.04)(1.11)

109

10

1,181,161,18

3.4761.8333.773

0.07860.19460.0679

SeptemberWhite shrimpBlue crabDaggerblade grass shrimp

22.016.817.5

(1.75)(1.70)(0.98)

101010

33.412.220.7

(2.14)(0.42)(2.05)

101010

1,181,181,18

17.3126.9491.989

0.0006*0.01680.1755

(ANOVA contrast, d.f.51,55, F50.921, p50.3413) orterrace pond (mean521.1 mm) sites (ANOVA con-trast, d.f.51,55, F50.376, p50.5424). Blue crab weresignificantly larger at terrace marsh than referencepond sites (means: 16.8 mm vs 8.7 mm, ANOVA con-trast, d.f.51,48, F524.280, p50.0001), although notsignificantly different between terrace pond(mean510.4 mm) and reference pond sites (ANOVAcontrast, d.f.51,48, F51.367, p50.2480).

Submerged aquatic vegetation was not widely dis-

tributed in the study area, but Ruppia maritima waspresent in some terrace ponds and in the referencepond. Based on the occurrence of SAV at our samplesites, we estimated the portion of total bottom areaoccupied by Ruppia to be 0.1% (May) and 4.8% (Sep-tember) in the terrace ponds and 0.2% (May) and 0.3%(September) in the reference pond. Even when mostabundant in September, the distribution of Ruppiawithin the terrace ponds was very patchy. BecauseSAV was present in so few samples, we did not ana-lyze these data further.

We found measurable differences in environmentalconditions among some of the treatment areas (Table5). Reference pond sites generally had higher dis-solved oxygen concentrations, deeper water, and werelocated farther from the shoreline than terrace pond orterrace marsh sites (Table 5). In contrast, terrace marshand reference marsh sites had similar environmentalcharacteristics; none of the means of the variables wemeasured (except salinity in May) were significantlydifferent between the two marsh treatments (Table 5).

DISCUSSION

The marsh terraces at Sabine NWR provided habitatsupport for some species that was equivalent to naturalareas. In our study, terrace marsh supported densitiesof white shrimp similar to those of a natural marsh,and both white shrimp and gulf menhaden were moreabundant within terrace ponds than in a nearby naturalpond. Moreover, the abundance and biomass of spe-cies that use the marsh surface were generally muchgreater in terrace marsh than over non-vegetated pondbottom; a terrace field would support higher standingcrops (numbers and biomass) of most of these speciesthan a shallow marsh pond of similar size. When whiteshrimp, brown shrimp, and blue crab were abundantin the study area, their mean densities were 8.0, 2.9,


Tabl

e5.

Env

iron

men

tal

char

acte

rist

ics

oftr

eatm

ent

area

s.M

ean

and

(1SE

)ar

egi

ven

for

vari

able

sm

easu

red

inm

arsh

vege

tatio

nan

dov

erno

n-ve

geta

ted

botto

mat

terr

ace

and

refe

renc

ear

eas

sam

pled

inM

ayan

dSe

ptem

ber

1999

.E

ach

mea

nis

estim

ated

from

10m

arsh

site

sor

20sa

mpl

esta

ken

inte

rrac

eor

refe

renc

epo

nds

(exc

ept

Oxy

gen

inM

ay:

Terr

ace

mar

sh5

6,R

efer

ence

mar

sh5

4,Te

rrac

epo

nd5

9,an

dR

efer

ence

pond

59)

.R

esul

ts(p

valu

es)

are

give

nfo

rA

NO

VA

anal

yses

we

used

toco

mpa

rem

eans

amon

gth

efo

urtr

eatm

ent

area

s(e

xcep

tco

mpa

riso

nsfo

rst

emde

nsity

are

betw

een

terr

ace

mar

shan

dre

fere

nce

mar

shon

ly)

and

apr

iori

cont

rast

ste

stin

gfo

rsi

gnifi

cant

diff

eren

ces

betw

een:

15

terr

ace

mar

shan

dre

fere

nce

pond

;2

5te

rrac

epo

ndan

dre

fere

nce

pond

;an

d3

5te

rrac

em

arsh

and

refe

renc

em

arsh

.T

heA

NO

VA

mod

elus

edto

doth

ean

alys

esis

show

nin

Tabl

e1.

An

*in

dica

tes

prob

abili

tyva

lue

was

sign

ifica

ntfo

rth

eov

eral

lA

NO

VA

afte

ral

pha

was

adju

sted

asde

scri

bed

byR

ice

(198

9).

Var

iabl

e

Terr

ace

Mar

sh

Mea

nSE

Ref

eren

ceM

arsh

Mea

nSE

Terr

ace

Pond

Mea

nSE

Ref

eren

cePo

nd

Mea

nSE

AN

OV

Ap

Val

ue

Con

tras

tp

Val

ues

(1)

TM

arsh

vsR

Pond

(2)

TPo

ndvs

RPo

nd(3

)T

Mar

shvs

RM

arsh

May

SAL

INIT

Y(‰

)O

XY

GE

N(p

pm)

WA

TE

RT

EM

PER

AT

UR

E(8

C)

16.8 4.8

24.4

(0.1

3)(0

.64)

(0.5

2)

15.4 5.0

25.5

(0.4

5)(0

.52)

(0.6

6)

16.8 5.1

24.8

(0.1

4)(0

.37)

(0.3

9)

15.8 6.8

24.6

(0.1

6)(0

.04)

(0.3

9)

0.00

01*

0.00

19*

0.54

85

0.00

160.

0011

0.00

030.

0012

0.00

030.

7851

WA

TE

RD

EPT

H(c

m)

TU

RB

IDIT

Y(F

TU

)D

IST

AN

CE

TO

ED

GE

(m)

STE

MD

EN

SIT

Y(s

tem

sm

22 )

33.3

48.0 0.7

74

(2.1

7)(1

1.25

)(0

.06)

(11.

2)

33.6

39.5 0.8

66

(2.3

5)(1

4.99

)(0

.08)

(8.0

)

46.7

34.4

10.5

(1.0

0)(6

.96)

(1.5

0)

56.6

68.8

23.3

(2.3

8)(1

2.27

)(2

.99)

0.00

01*

0.09

060.

0001

*0.

5580

0.00

01

0.00

01

0.00

02

0.00

01

0.93

22

0.98

17

Sept

embe

rSA

LIN

ITY

(‰)

OX

YG

EN

(ppm

)W

AT

ER

TE

MPE

RA

TU

RE

(8C

)

26.0 5.8

30.6

(0.3

7)(0

.56)

(0.4

2)

27.4 7.2

31.2

(0.4

3)(0

.38)

(0.3

5)

25.9 6.0

31.5

(0.2

4)(0

.45)

(0.6

7)

26.8 7.7

32.2

(0.3

5)(0

.20)

(0.3

6)

0.01

390.

0029

*0.

2478

0.14

620.

0036

0.03

460.

0013

0.02

060.

0590

WA

TE

RD

EPT

H(c

m)

TU

RB

IDIT

Y(F

TU

)D

IST

AN

CE

TO

ED

GE

(m)

STE

MD

EN

SIT

Y(s

tem

sm

22 )

15.3

32.1 0.1

110

(2.1

5)(1

2.87

)(0

.10)

(13.

9)

18.2

28.6 0.2

101

(1.5

3)(1

3.09

)(0

.11)

(20.

7)

22.4

16.8

11.1

(1.0

9)(2

.35)

(1.1

1)

36.3

15.7

23.7

(1.0

4)(1

.90)

(3.1

9)

0.00

01*

0.22

320.

0001

*0.

7316

0.00

01

0.00

01

0.00

01

0.00

01

0.21

43

0.97

98


and 2.4 times greater, respectively, in terrace marshvegetation than in the reference pond. Because ourstudy only examined one terrace field in comparisonto one reference pond, it was not possible to examinevariability among terrace fields and reference ponds.This problem with pseudoreplication (Hurlbert 1984)is not uncommon in assessments of created wetlands(Minello and Zimmerman 1992, Streever 2000). How-ever, as documented in numerous studies elsewhere(Zimmerman and Minello 1984, Minello et al. 1994,Cicchetti 1998, Howe et al. 1999, Minello 1999, Rozasand Zimmerman 2000), penaeid shrimps and blue crabare strongly attracted to shoreline vegetation. Most ofthese species should clearly benefit by replacing someshallow, non-vegetated bottom with marsh terraces inestuaries where the area of emergent marsh has de-creased and restoration by marsh terracing is practical.

Marsh terraces constructed in non-vegetated areasshould provide animals with protection from largepredators by adding vegetation structure in the formof emergent marsh on terrace ridges and, perhaps, byfacilitating the growth of SAV within terrace ponds.Both emergent vegetation and submerged aquaticshave been shown to reduce predation rates on speciescollected in our study (Heck and Thoman 1981, Mi-nello and Zimmerman 1983, Wilson et al. 1987, 1990,Rozas and Odum 1988, Minello et al. 1989, Minello1993). Although SAV was not widely distributed with-in the terrace ponds when we sampled the sites at Sa-bine NWR, Ruppia may be more widespread there atother times.

Local sediment processes also may be affected bythe presence of terrace fields. Initial data collected atSabine NWR indicated that turbidity was reducedwithin the terrace ponds compared with surroundingopen-water areas (Underwood et al. 1991); results ofextended monitoring, however, did not show a signif-icant reduction in turbidity levels within the terraceponds (LDNR 1993). In our study, the mean turbiditylevel in terrace ponds was about half that of the ref-erence pond in May, but turbidity levels were not sig-nificantly different among treatments. In September,turbidity levels in ponds and the reference area werelow and very similar. Some evidence for accretionwithin terrace ponds was reported initially, but sub-sequent and more-detailed analyses were inconclusivebecause most of the feldspar plots placed in the areato measure sediment accretion could not be located(LDNR 1993). Accretion rates do not appear to havebeen high enough to facilitate expansion of Spartinaacross the terrace ponds, and we saw no evidence thatemergent vegetation will eventually cover the ponds.Initial monitoring of the Sabine NWR terracing didshow that the terraces reduced fetch, wave height, and

shoreline retreat within the terraced marsh ponds(LDNR 1993).

The Spartina marsh established on terrace levees atSabine NWR was not functionally equivalent to anadjacent natural marsh, even after 9 years of devel-opment. In our study, the terrace marsh supportedbrown shrimp, blue crab, daggerblade grass shrimp (inSeptember), and brackish grass shrimp at densitiesabout half or less those of the nearby natural marsh.Moreover, more striped mullet (May), white shrimp(September), and spotted seatrout (September) bio-mass was present in the natural marsh than in the ter-race marsh. Similarly, densities of daggerblade grassshrimp and juvenile brown shrimp were significantlyless in marshes constructed on dredged material thanin natural marshes in Texas (Minello and Zimmerman1992). In a more extensive study of 10 created marshesand five natural marshes in Galveston Bay, Minelloand Webb (1997) reported that densities of brownshrimp, blue crab, and white shrimp in created marsheswere only 25–40%, 5–56%, and 59%, respectively, ofthose in natural marshes. Interestingly, when they re-peated the comparison for white shrimp after removingdata from created marshes with relatively high surfaceelevations and low flooding duration, densities be-tween created marshes and natural marshes were notsignificantly different (Minello and Webb 1997). Thischange did not occur for brown shrimp and blue crab;densities of these two species were still significantlygreater in natural marshes even when the high-eleva-tion created marshes were removed from the analysis.Perhaps white shrimp will use created marsh as readilyas natural marsh when the marsh is constructed at anelevation similar to natural marsh. Results from otherstudies suggest that marsh surface elevation can influ-ence habitat use by white shrimp and other species(Rozas and Reed 1993, Rozas and Zimmerman 2000).In our study, terrace marsh and reference marsh siteswere similar in mean water depth, and mean densitiesof white shrimp in these two habitat types were nearlyequal. Measures of water depth in our study are a goodproxy for relative elevation because tidal fluctuationswere low and our samples of the two marsh types wereinterspersed in time.

Comparisons of animal size between natural andcreated marshes may indicate differences in growthrates and habitat quality. Minello and Webb (1997)suggested that grass shrimp may have been smaller increated marshes due to inadequate food supply in thesemarshes. In our study, we detected few differences inthe mean size of animals between the terrace marshand the reference marsh. White shrimp, however, weresignificantly smaller in terrace marsh than naturalmarsh.

The terrace marsh was about 9 years old when we


collected our samples, but there is some evidence thatage of created marshes has little relationship with hab-itat value for nekton. Recent studies in Galveston Bay,Texas showed that marsh established on dredged ma-terial develops rapidly and apparently reaches its max-imum habitat support function for nekton in just overa year following construction (Goldberg 1996, Minello2000). For some species, however, this level of habitatsupport may never reach equivalency with naturalmarshes (Minello and Webb 1997).

Design changes to increase the relative area ofmarsh in a terrace field should be able to enhancestanding crops of species that select for marsh habitat.As an example, we simulated the effect of increasingmarsh area on the standing crop of brown shrimp in aterrace field. Increasing the area of marsh within theterrace field from 21 to 26% would boost the standingcrop of brown shrimp in spring, making it equivalentto the marsh pond standing crop. Increasing the per-centage of marsh in the terrace field to 35% wouldincrease the standing crop to 20,000 brown shrimpha21, a 25% increase over the level in a marsh pond.The area of marsh could be enlarged by increasing thedensity of terrace ridges and reducing the size of cellsin a terrace field. However, other design changes alsoare possible. For example, terrace ridges can be laidout in various patterns other than the checkerboardform. Design considerations for maximizing the hab-itat function of marsh terraces should be assessed infuture research.

Interestingly, the white shrimp densities we ob-served in our study are some of the highest reportedin coastal environments. White shrimp occurred atmean densities of . 60 m22 in both of our marsh treat-ments, and four of our samples within Spartina veg-etation contained . 100 individuals m22. Other studiesof shoreline Spartina marsh using similar gear andmethods have reported mean white shrimp densities ashigh as 49.6 m22 (Rozas and Minello 2000) and 30.0m22 (Minello and Webb 1997) in Galveston Bay, Tex-as. Lower densities were reported for Mobile Bay (1.4m22, Howe et al. 1999) and other locations in Texasand Louisiana (1.6–26.7 m22, Zimmerman et al. 2000).Densities of other decapod crustaceans (brown shrimp,pink shrimp, blue crab, grass shrimp) collected in ourstudy were similar to those found in other studies ofsimilar habitat.

In summary, marsh terracing appears to reducefetch, wave energy, and shoreline erosion within coast-al water bodies. Terracing also may provide favorableconditions within ponds for SAV establishment andgrowth, but we found no evidence for increased SAVcover in the terrace fields at Sabine NWR. For manynekton species, marsh terracing appears to increasestanding crop measured in both numbers and biomass

per hectare. Densities of white shrimp, brown shrimp(spring), and blue crab were much greater in terracemarsh than over non-vegetated mud bottom, the hab-itat type that is replaced when marsh-terracing projectsare constructed. The terrace marsh at Sabine NWR,however, was not functionally equivalent to nearbynatural marsh based on densities of brown shrimp,blue crab, and grass shrimps or biomass of whiteshrimp, striped mullet, and spotted seatrout. Future ter-racing plans could enhance the habitat value of pro-jects for fishery species by increasing the proportionof marsh in terrace fields. We believe that marsh ter-racing can be used successfully as a restoration tech-nique to enhance coastal fisheries.

ACKNOWLEDGMENTS

This research was conducted through the FisheryEcology Branch (FEB) of the Galveston Laboratory inthe Southeast Fisheries Science Center (NOAA, Na-tional Marine Fisheries Service). The assistance of ev-eryone in the FEB was essential for the successfulcompletion of this project. In particular, we thankHeather Carter, Jim Ditty, Jennifer Doerr, DavidEvans, Kirk Kilfoyle, Seth King, and Shawn Hillen forhelping collect and process samples and Phillip Cald-well for producing Figure 2. We also thank ParcyAbohweyere (Nigerian Institute of Oceanography andMarine Research), Cherie O’Brien (Texas Parks andWildlife Department), and Bill Streever (U.S. ArmyCorps of Engineer Waterways Experiment Station),who volunteered to help us collect the nekton samples.We gratefully acknowledge the logistical support andinformation about the Sabine National Wildlife Refugeprovided by Chris Pease, Herb Bell, Ben Mense, RoyWalker, and other refuge personnel. Thanks to RonBoustany and Tommy Michot (USGS, National Wet-lands Research Center) for providing the photographfor Figure 1. The suggestions of Bill Good, WilliamJ. Richards, Greg D. Steyer, Bill Streever, and twoanonymous reviewers improved this paper. We ac-knowledge the U.S. Army Corps of Engineer Water-ways Experiment Station and the National MarineFisheries Service for funding this project.

LITERATURE CITED

Baltz, D. M., C. Rakocinski, and J. W. Fleeger. 1993. Microhabitatuse by marsh-edge fishes in a Louisiana estuary. EnvironmentalBiology of Fishes 36:109–126.

Boyer, M. E. 1997. The effect of long-term marsh management onland-loss rates in coastal Louisiana. Environmental Management21:97–104.

Chamberlain, R. H. and R. A. Barnhart. 1993. Early use by fish ofa mitigation salt marsh, Humboldt Bay, California. Estuaries 16:769–783.

Cicchetti, G. 1998. Habitat use, secondary production, and trophic ex-


port by salt marsh nekton in shallow water. Ph.D. Dissertation. TheCollege of William and Mary in Virginia, Williamsburg, VA, USA.

Delaney, T. P., J. W. Webb, and T. J. Minello. 2000. Comparisonof physical characteristics between created and natural estuarinemarshes in Galveston Bay, Texas. Wetlands Ecology and Man-agement 8:343–352.

Gilmore, R. G., D. W. Cooke, and C. J. Donohoe. 1982. A com-parison of the fish populations and habitat in open and closed saltmarsh impoundments in east-central Florida. Northeast Gulf Sci-ence 5:25–37.

Goldberg, A. R. 1996. Development of infaunal populations andbelow-ground organic matter from three created Spartina alter-niflora marshes in Galveston Bay, Texas. M.S. Thesis. TexasA&M University, College Station, TX, USA.

Heck, K. L. and T. A. Thoman. 1981. Experiments on predator-preyinteractions in vegetated aquatic habitats. Journal of ExperimentalMarine Biology and Ecology 53:125–134.

Howe, J. C., R. K. Wallace, and F. S. Rikard. 1999. Habitat utili-zation by postlarval and juvenile penaeid shrimps in Mobile Bay,Alabama. Estuaries 22:971–979.

Hurlbert, S. H. 1984. Pseudoreplication and the design of ecologicalfield experiments. Ecological Monographs 54:187–211.

Kamps, L. F. 1962. Mud Distribution and Land Reclamation in theEastern Wadden Shallows. Rijkswaterstaad Communications, TheHague, The Netherlands.

Kneib, R. T. 1997. The role of tidal marshes in the ecology ofestuarine nekton. Oceanography and Marine Biology: an AnnualReview 35:163–220.

LDNR. 1993. Sabine Terracing Project Final Report. Coastal Res-toration Division, Louisiana Department of Natural Resources,Baton Rouge, LA, USA. DNR Project No. 4351089.

McGovern, J. C. and C. A. Wenner. 1990. Seasonal recruitment oflarval and juvenile fishes into impounded and non-impoundedmarshes. Wetlands 10:203–221.

Meyer, D. L., M. S. Fonseca, V. G. Thayer, W. J. Kenworthy, andD. R. Colby. 1993. Faunal use of restored marsh and seagrasshabitat. p. 1858–1863. In O.T. Magoon, H. Converse, D. Miner,L. T. Tobin, and D. Clark (eds.) Coastal Zone ’93, Vol 4. Pro-ceedings of the 8th symposium on coastal ocean management.American Society of Civil Engineers, New York, NY, USA.

Minello, T. J. 1993. Chronographic tethering: a technique for mea-suring prey survival time and testing predation pressure in aquatichabitats. Marine Ecology Progress Series 101:99–104.

Minello, T. J. 1999. Nekton densities in shallow estuarine habitatsof Texas and Louisiana and the identification of Essential FishHabitat. p. 43–75. In L. Benaka (ed.) Fish Habitat: Essential FishHabitat and Habitat Rehabilitation. American Fisheries SocietySymposium 22, Bethesda, MD, USA.

Minello, T. J. 2000. Temporal development of salt marsh value fornekton and epifauna: utilization of dredged material marshes in Gal-veston Bay, Texas. Wetlands Ecology and Management 8:327–341.

Minello, T. J. and J. W. Webb Jr. 1997. Use of natural and createdSpartina alterniflora salt marshes by fishery species and otheraquatic fauna in Galveston Bay, Texas, USA. Marine EcologyProgress Series 151:165–179.

Minello, T. J. and R. J. Zimmerman. 1983. Fish predation on ju-venile brown shrimp, Penaeus aztecus Ives: the effect of simulatedSpartina structure on predation rates. Journal of Experimental Ma-rine Biology and Ecology 72:211–231.

Minello, T. J. and R. J. Zimmerman. 1992. Utilization of naturaland transplanted Texas salt marshes by fish and decapod crusta-ceans. Marine Ecology Progress Series 90:273–285.

Minello, T. J., R. J. Zimmerman, and E. X. Martinez. 1989. Mor-tality of young brown shrimp Penaeus aztecus in estuarine nurs-eries. Transactions of the American Fisheries Society 118:693–708.

Minello, T. J., R. J. Zimmerman, and R. Medina. 1994. The impor-tance of edge for natant macrofauna in a created salt marsh. Wet-lands 14:184–198.

Mitsch, W. J. and J. G. Gosselink. 1986. Wetlands. Van NostrandReinhold Company Inc., New York, NY, USA.

Moy, L. D. and L. A. Levin. 1991. Are Spartina marshes a replace-able resource? A functional approach to evaluation of marsh cre-ation efforts. Estuaries 14:1–16.

Orlando, S. P. Jr, L. P. Rozas, G. H. Ward, and C. J. Klein. 1993.Salinity characteristics of Gulf of Mexico estuaries. NOAA Officeof Ocean Resources Conservation and Assessment, Silver Spring,MD, USA.

Peck, M. A., P. E. Fell, E. A. Allen, J. A. Gieg, C. R. Guthke, andM. D. Newkirk. 1994. Evaluation of tidal marsh restoration: com-parison of selected macroinvertebrate populations on a restoredimpounded valley marsh and an unimpounded valley marsh withinthe same salt marsh system in Connecticut, USA. EnvironmentalManagement 18:283–293.

Perez-Farfante, I. and B. Kensley. 1997. Penaeoid and sergestoidshrimps and prawns of the world: Keys and diagnoses for thefamilies and genera. Memoires du Museum National d’HistorieNaturelle, Tome 175, Paris, France.

Rey, J. R., J. Shaffer, D. Tremain, R. A. Crossman, and T. Kain.1990. Effects of re-establishing tidal connections in two impound-ed subtropical marshes on fishes and physical conditions. Wet-lands 10:27–45.

Rice, W. R. 1989. Analyzing tables of statistical tests. Evolution 43:223–225.

Rogers, D. R., B. D. Rogers, and W. H. Herke. 1994. Structuralmarsh management effects on coastal fishes and crustaceans. En-vironmental Management 18:351–369.

Rozas, L. P. and T. J. Minello. 1997. Estimating densities of smallfishes and decapod crustaceans in shallow estuarine habitats: areview of sampling design with focus on gear selection. Estuaries20:199–213.

Rozas, L. P. and T. J. Minello. 1998. Nekton use of salt marsh,seagrass, and nonvegetated habitats in a South Texas (USA) es-tuary. Bulletin of Marine Science 63:481–501.

Rozas, L. P. and T. J. Minello. 1999. Effects of structural marsh man-agement on fishery species and other nekton before and during aspring drawdown. Wetlands Ecology and Management 7:121–139.

Rozas, L. P. and W. E. Odum. 1988. Occupation of submergedaquatic vegetation by fishes: testing the roles of food and refuge.Oecologia 77:101–106.

Rozas, L. P. and D. J. Reed. 1993. Nekton use of marsh-surfacehabitats in Louisiana (USA) deltaic salt marshes undergoing sub-mergence. Marine Ecology Progress Series 96:147–157.

Rozas, L. P. and R. J. Zimmerman. 2000. Small-scale patterns ofnekton use among marsh and adjacent shallow nonvegetated areasof the Galveston Bay Estuary, Texas (USA). Marine Ecology Pro-gress Series 193:217–239.

Streever, W. 2000. Spartina alterniflora marshes on dredged mate-rial: a critical review of the ongoing debate over success. WetlandsEcology and Management 8:295–316.

Underwood, S. G., G. D. Steyer, B. Good, and D. Chambers. 1991.Bay bottom terracing and vegetative planting: an innovative ap-proach for habitat and water quality enhancement. p. 164–173. InF. J. Webb, Jr. (ed.) Annual Conference on Wetlands Restorationand Creation. Hillsborough Community College, Tampa, FL, USA.

U.S. Department of Commerce. 1998. Tide tables 1999 (high and lowwater predictions) east coast of North and South America, includingGreenland. NOAA National Ocean Service, Rockville, MD, USA.

Wilson, K. A., K. W. Able, and K. L. Heck. 1990. Predation rateson juvenile blue crabs in estuarine nursery habitats: evidence forthe importance of macroalgae (Ulva lactuca). Marine EcologyProgress Series 58:243–251.

Wilson, K. A., K. L. Heck, and K. W. Able. 1987. Juvenile bluecrab, Callinectes sapidus, survival: an evaluation of eelgrass, Zos-tera marina, as refuge. Fishery Bulletin, US 85:53–58.

Zimmerman, R. J. and T. J. Minello. 1984. Densities of Penaeusaztecus, Penaeus setiferus, and other natant macrofauna in a Texassalt marsh. Estuaries 7:421–433.

Zimmerman, R. J., T. J. Minello, and L. P. Rozas. 2000. Salt marshlinkages to productivity of penaeid shrimps and blue crabs in thenorthern Gulf of Mexico. p. 293–314. In M. P. Weinstein and D.A. Kreeger (eds.) Concepts and controversies in tidal marsh ecol-ogy. Kluwer Academic Publishers, The Netherlands.

Zimmerman, R. J., T. J. Minello, and G. Zamora. 1984. Selectionof vegetated habitat by brown shrimp, Penaeus aztecus, in a Gal-veston Bay salt marsh. Fishery Bulletin, US 82:325–336.

Manuscript received 10 October 2000; revisions received 12 March2001; accepted 20 April 2001.

marsh terracing as a wetland …courses.washington.edu/esrm479/marsh terracing.pdf327 wetlands, vol....

Documents