ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005 235

Ecological Restoration, Vol. 23, No. 4, 2005 ISSN 1522-4740 E-ISSN 1543-4079©2005 by the Board of Regents of the University of Wisconsin System.

Much time and many resources arespent managing coastal wetlands forthe roughly four million waterfowl thatwinter in Louisiana marshes (Boesch andothers 1994, Michot 1996). Waterfowlmanagers in coastal Louisiana often aimto increase submergent aquatic vegeta-tion (SAV) abundance to support morewaterfowl because duck populations inwinter are believed to be food limited(Esslinger and Wilson 2001). Manage-ment of marshes with fixed-crest weirsincreases SAV by minimizing fluctuationsin salinity, water levels, and turbidity(Larrick and Chabreck 1976), but theeffect of weirs varies from year to year(Nyman and Chabreck 1996).

Joanen and Glasgow (1965) reporteda winter decline in Louisiana SAV—declines that may be caused by winteringwaterfowl depleting SAV through feed-ing, or in response to seasonal changessuch as water temperature or day length(Pulich, Jr. 1985). However, there are nodata, other than that collected by Joanenand Glasgow (1965), relating SAV abun-dance on the Gulf Coast to waterfowl her-bivory, and those results never have beenreplicated. Our objective was to docu-ment the effects of season and a new typeof management on SAV abundance inorder to determine what seasonal patterns

exist, if any, and to determine if manage-ment was effective at increasing SAV. Todetermine why management was effec-tive or ineffective, we also documentedwater quality parameters (salinity, tem-perature, water level, turbidity, and nutri-ents) and tested a second hypothesis inwhich we postulated that these parame-ters did not differ among the study areas orwith SAV abundance.

Study LocationWe conducted the study at Marsh Island,which lies on the coast of Louisianabetween Vermilion Bay and the Gulf ofMexico (Figure 1). There are 76,570 acres(31,000 ha) of tidally influenced brackishmarsh and interior ponds on the island. In1920, the Russell Sage Foundationdonated the island to Louisiana with theunderstanding that the state would man-age it for wintering waterfowl in particularand wildlife in general. Shallow ponds onthe island historically have served as feed-ing ground for millions of waterfowl eachyear (Louisiana Department of Wildlifeand Fisheries, unpublished letter fromMcIlhenny and others to Viosca 1934).

Emergent vegetation at Marsh Islandis dominated by saltmeadow cordgrass(Spartina patens). A study by Nyman and

Effects of Season and MarshManagement on SubmergedAquatic Vegetation inCoastal Louisiana BrackishMarsh Pondsby Joy H. Merino, John A. Nyman and Thomas Michot

RESEARCH

Correct water depth is

critical for increasing

submerged aquatic

vegetation in man-

aged marshes along

the Louisiana coast.

Keywords: marsh management, Ruppia mar-itima, submerged aquatic vegetation, coastalecosystems, waterfowl management

Chabreck (1996) found that the four mostabundant SAV species in order of abun-dance are: watermilfoil (Myriophyllum spi-catum), widgeon grass (Ruppia maritima),wild celery (Vallisneria americana), andcoontail (Ceratophyllum demersum).

Marsh ManagementAlthough water levels are unmanaged onabout 63,000 acres (25,500 ha) of MarshIsland, several large tracts are subject toactive or passive water-level management.The largest managed area is a 7,904-acre(3,200-ha) impoundment constructed in1958. About 3,700 acres (1,500 ha) aremanaged passively through the use offixed-crest weirs that were also con-structed in 1958 (Louisiana Wild Life andFisheries Commission 1959). Fixed-crestweirs prevent ponds from draining morethan 6 inches (15 cm) below marsh ele-vation during the low-water periods thatoccur periodically during winter. Previousstudies have found that areas managedwith fixed-crest weirs generally supportmore SAV than non-managed areas onthe island (Nyman and Chabreck 1996),and that waterfowl are more abundant in

weir-managed areas than in non-managedareas (Spiller and Chabreck 1975). Thereare concerns, however, that fixed-crestweirs reduce abundance of estuarine-dependent fish and crustaceans (Hoeseand Konikoff 1995).

In 1993, the Louisiana Departmentof Natural Resources constructed a pro-ject, known as, TV-06 Marsh IslandControl Structures (LDNR 1994). In thatproject, they used water control structuresto actively manage two 988-acre (400-ha)management units. These units haveidentical structures, flap-gated culvertswith variable-crest weirs, in the spoilbanks that separate the units from canalsdredged in the 1950s (LDNR 1994). Thecrest height determines the minimumwater level in ponds. Flap-gates allowwater to drain out of the units during lowtides, whereas they restrict inflow duringmost high tides (Figure 2). Unlike tradi-tional marsh management units through-out the southeastern United States, thesestwo units lack levees. The lack of leveesallows water entry into the units duringperiods of higher water associated withoccasional normal tides and frequent win-ter and tropical storms. Variable-crest

weirs combined with flap-gated culvertsallow managers to drain ponds (low crest,flaps down), hold water in ponds (crest 1ft or 30 cm below marsh elevation, flapsup or down), or maximize water exchange(low crest, flaps up).

Project GoalsThe goals of the TV-06 project were toimprove the habitat by reducing land lossand to promote submerged and emergentaquatic vegetation in the two units(LDNR 1994). This project containedrestoration and management aspects. Itcould be classified as restoration because itwas designed to reduce tidal action ininterior marsh ponds where tidalexchange had been increased by a naviga-tion channel that was created during the1950s to accommodate oil and gas explo-ration. It could be classified as manage-ment because it was designed to allowartificial drawdowns of interior marshponds. Throughout the rest of the article,we refer to the areas as management areasrather than restoration areas.

Hydrologic goals of management inthese two study areas are 1) to reduce salt-water stress on marsh vegetation, 2) toretain water in the ponds for waterfowlduring fall and spring, and 3) to facilitatethe removal of water from the ponds dur-ing spring and summer to encourage plantgermination and growth. Marsh managersassume that periodic removal of waterfrom ponds (drawdown) promotes SAV bystabilizing sediments and thereby reducingturbidity (Chabreck 1994). McGinnis(1997) found that this managementapproach reduced marsh erosion rates inponds smaller than 32.8 ft (10 m) in diam-eter within these two areas, but he did notstudy the effect on SAV abundance.

Experimental DesignWe studied the two units constructed forthe TV-06 project and two unmanagedareas (Figure 1). We refer to these areas as:North Managed (NM), North Unman-aged (NUM), South Managed (SM), andSouth Unmanaged (SUM). Each man-aged area has a flap-gated, variable-crestweir that separates the areas from the

236 ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005

Figure 1. Study areas at Marsh Island, Louisiana. NM = North Managed, NUM = North

Unmanaged, SM = South Managed, SUM = South Unmanaged.

channels that were artificially deepenedin the 1950s to accommodate navigation.

We found noticeable geomorphic dif-ferences throughout the landscape. Pondsin the north areas were more uniformlyshaped (round) than ponds in the southareas. We accounted for such landscapevariability in the experimental design bytesting for differences among all four areasrather than assuming the two unmanagedareas were similar and the two managedareas were similar. Although this approachassumed that SAV abundance was similaramong the four areas prior to manage-ment, testing for differences among allfour areas allowed us to check, rather thanassume, that the two unmanaged areaswere similar and the two managed areaswere similar during the study.

There are records for SAV abundanceelsewhere on the island beginning in the1950s (Nyman and Chabreck 1996), butnot from our study areas. However, Orton(1959) made earlier observations thatindicated no differences in emergent veg-etation or degree of drainage among thestudy areas at Marsh Island. All areas wereclassified as having poorly drained soilsand black rush-big cordgrass (Juncus rome-rianus-Spartina cynosuroides) vegetation.We, therefore, assumed that differencesbetween managed and unmanaged areasresulted from management. If thisassumption is false, then conclusionsabout the effects of management are like-wise faulty.

Depth of ponds in all areas rangedfrom roughly 0.4 to 12 inches (1 to 30 cm)below the elevation of the adjacent marsh(McGinnis 1997). Pond depth did not dif-fer between the NM and NUM, or betweenthe SM and SUM (McGinnis 1997).

Within each of the four study areas,we numbered all ponds measuringbetween 115 and 541 ft (35 and 165 m) indiameter with the aid of aerial pho-tographs. We then randomly selected sixnumbered ponds in each area. From thosesix ponds, we used three for water and soilcharacteristics and three for SAV abun-dance studies in each area.

Measuring SAVAbundance, WaterParameters, and SedimentCharacteristicsSAV AbundanceWe measured SAV abundance every six toeight weeks between October 1998 andMay 2000 by making estimates of biomassand percent cover. We used these twomeasures because neither is effectivealone across a wide range of abundance.While biomass estimates are preferred,estimating the biomass of ponds is ineffi-cient when SAV beds are small and few.At the other end of the spectrum, esti-mating the percent of pond bottom cov-ered by SAV fails to provide meaningfulinformation when SAV beds cover virtu-ally the entire pond bottom. Estimatingboth biomass and percent cover was animprovement over previous reports thathave addressed temporal variation inabundance by sampling biomass onlywhen it was at high levels (Fores-Verdugoand others 1988).

To estimate biomass per square meter,we collected 10-cm-diameter cores fromfive randomly selected points in eachpond. We used the cores to extract all

vegetation in the water column withinthe upward projection of the sampledarea, and all sediment and root materialdown to a depth of at least 11.7 inches (30cm). We kept all material retained by a 2-mm sieve (Ellison and others 1986). Ourmethod also allowed us to estimate thebiomass of SAV per square meter of a bedby ignoring cores lacking SAV.

In the lab, technicians separated SAVby species into live aboveground and livebelowground biomass. Samples wereplaced on pre-weighed foil sheets and driedin a drying oven at 178˚F (81˚C) for a min-imum of 72 hours. During data analysis,live aboveground and live belowgroundparts were pooled for all species except wid-geon grass, which had easily distinguish-able belowground tissues even when notintact. Other species’ belowground tissueswere visually indistinguishable.

We estimated percent SAV coverage(percent of pond bottom covered withSAV) using a sampling method describedby Nyman and Chabreck (1996). Thisinvolved standing in an idling airboat andtaking a grab sample from the pond bot-tom with a garden rake about every 10 ftor 3 m along two transects spaced roughlyone-third the pond width apart. Therewere at least 20 grab samples per transect,

ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005 237

Figure 2. Flap-gated culverts with variable–crest weirs in the spoil banks that separate man-

aged marsh units from canals. The managed units lack levees, allowing water entry into the

units during periods of higher water associated with occasional normal tides and frequent win-

ter and tropical storms. Photo courtesy of Joy Merino

with larger ponds resulting in more sam-ples. Cover of SAV on each transect wascalculated as the number of times SAVwas present divided by the number oftimes the rake touched the pond bottom.Percent cover of SAV in the pond wasestimated from the average of the twotransects in the pond. We did not usevisual estimates of cover because pres-ence/absence data provides a better esti-mate of cover than visual estimates,which are subjective (Kershaw andLooney 1985, Moore and Chapman 1986,Bonham 1989), and because turbiditygenerally was great enough to obscureSAV even when SAV almost coveredpond bottoms.

Because grab samples are equivalentto pins and the transect equivalent to apin frame (Kershaw and Looney 1985,Causton 1988), the resulting percentcover estimates were biased upwards, asoccurs with all pin or plot methods(Grieg-Smith 1964, Kershaw and Looney1985, Moore and Chapman 1986,Bonham 1989). In fact, our techniquebiased cover upward more than pinswould have because our rake grabbed froma 0.02-m2 area. Nevertheless, they repre-sent a valid estimate of the percent of thepond bottom covered by SAV bedsbecause our plot size was similar to thatrecommended by Daubenmire (1968) forsuch a sampling method.

Although dwarf spikerush (Eleocharisparvula) and bacopa (Bacopa monneri) arenot SAV, we included these species in theanalysis of SAV because they are potentialwaterfowl food sources and occurred sub-merged in shallow areas of ponds whereSAV grows.

Percent cover, biomass, and residualsof the model for percent cover and bio-mass were not normally distributed andvariances were unequal. Because transfor-mations of the data failed to normalize thedata and ranking of the residuals failed toimprove homogeneity of variance, statis-tical significance was determined with arandomization method described byAdams and Anthony (1996) to test thenull hypotheses that 1) biomass and per-cent cover did not vary with time, andthat 2) biomass and percent cover did notdiffer among the four areas.

Randomization is a method of resam-pling. This method compares the observedtreatment sum-of-squares (SS) for thedata to SS generated from observationsrandomly assigned to treatments. The fre-quency of observations in a treatment ismaintained, the observed data are ran-domly assigned to treatments, and SS iscalculated. The random assignment ofobservations to treatments and calcula-tion of the SS is repeated 5,000 times. Thefrequency of randomly generated SSgreater than or equal to the observed SS isan estimate of the likelihood of obtainingthe observed SS by chance. Therefore,data may be analyzed without any assump-tions of distribution. Using this method-ology, we generated SS from a two-wayANOVA with percent cover or biomass asthe dependant variable and time, areaand the interaction between area andtime as dependent variables. An alphalevel of 0.05 was used to reject the nullhypotheses. We obtained 95-percent con-fidence intervals by bootstrapping themeans (Sokal and Rohlf 1995).

Water and SedimentCharacteristicsWe acquired water level, salinity, andtemperature data from the LouisianaDepartment of Wildlife and Fisheries(LDWF) that they took from October1998 to May 2000. A Yellow SpringsInstruments (YSI) Model 6000 meter anddata-logger recorded conductivity, salin-ity, temperature, and water level in eachstudy area on an hourly basis. The sensorof the data logger was cleaned and cali-brated to standards every one to twomonths by LDWF staff. The instrumentdetermined salinity from conductivityand temperature, and water level from thewater pressure detected by a pressuretransducer. The LDWF staff used conven-tional survey techniques to determine ele-vation of the adjacent marsh surfacerelative to the depth sensor, and weexpressed all water levels as relative tolocal marsh surface elevation.

Unfiltered water samples were col-lected in 500-ml polyethylene bottlesfrom areas least disturbed by our boatactivity. We also took turbidity readings at

that time using a HANNA portable tur-bidity meter. The water samples werefrozen until they could be processed bytechnicians at the National WetlandsResearch Center. Samples were analyzedfor nitrite, nitrate, and soluble reactivephosphate (SRP) by an automated seg-mented flow analysis in an Alpkem FlowSolution III autoanalyzer. Ammoniumconcentrations were determined by col-orimetric analysis using an indophenolblue method (United States Environ-mental Protection Agency 1979). Dis-solved inorganic nitrogen (DIN) wascalculated as the sum of nitrate, nitrite,and ammonium readings.

We also collected a sample of the top4 inches (10 cm) of pond bottom sedi-ment from each pond. A representativesub-sample of sediment was dried in anoven at 177.8˚ (81˚C) to obtain dry mass.Dried samples were then ignited in a muf-fle furnace at 752˚ (400˚C) to obtain theash weight (mineral content). We calcu-lated the organic content by subtraction.

We decided not to use randomizationmethods for hourly recordings of salinity,temperature and water level because thesample size was too large (n = 58,560).Instead, we analyzed the daily means byrepeated measures under the split-plotframework with salinity, temperature,and water level as the dependent vari-ables and time, area, and the interactionas independent variables. We used analpha level of 0.05 to reject the nullhypotheses. Water nutrients and turbid-ity were analyzed by randomization asdescribed earlier.

We examined the relationshipbetween water quality variables and SAVabundance by using repeated measuresunder the split-plot framework with bio-mass and percent cover as dependent vari-ables and area, time, and the water qualityvariable of interest as independent vari-ables. Because SAV abundance and waterquality characteristics were sampled ondifferent dates, we created a new timevariable that condensed all dates towithin 30 days of one another. The meansof hourly samples (salinity, temperature,and water level relative to marsh surface)for 30 days prior to collection of SAVabundance samples were used for analysis.

238 ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005

There were not enough new variabledates to allow for a model simultaneouslyrelating all variables. Residuals of themodel with biomass and percent cover asdependent variables and area and time asindependent variables were analyzed forsignificant correlations with the waterquality variable of interest.

We decided to analyze SAV and waterquality characteristics by correlation with-out accounting for the effects of area anddate, in addition to repeated measuresdescribed above, because area and timecould be confounded with the effects ofwater quality characteristics. Moreover, ifmanagement was successful in alteringwater characteristics to promote SAV, thenwater characteristics would be confoundedwith area and date (management).

ResultsSAV AbundanceWe found that widgeon grass accountedfor 92 percent of all SAV biomass (Table1). Biomass of widgeon grass averaged 24g/m2 on a bed basis, but 7.6 g/m2 on apond basis. Biomass changed over timedifferently among the four areas (Parea *time = 0.01). Variability in biomass overtime was evident more in the NM than in

the other areas where biomass was gener-ally low and invariable (Figure 3).Biomass was greater in one or both man-aged areas than in the unmanaged areason five of the eight sample dates (Figure3). Submerged aquatic vegetation aver-aged 17.2 g/m2 in managed ponds, butonly 4.1 g/m2 in unmanaged ponds. Dif-ferences were also apparent between thetwo managed areas—the NM had morebiomass than the SM area on four of theeight sample dates (Figure 3). Biomass inthe unmanaged areas was never signifi-cantly greater than in the managed areas(Figure 3).

Widgeon grass was also the mostabundant SAV species observed when esti-mating percent cover (Table 1). Out of192 transects, 137 contained widgeongrass, while other SAV species were pre-sent in less than 23 transects. Cover ofwidgeon grass in ponds averaged 30 per-cent. Unvegetated bare sediment in pondsaveraged 64.7 percent. Like biomass, per-cent cover changed over time differentlyamong the four areas s (Parea * time =0.0236), but percent cover showed differ-ences among the areas not evident in thebiomass data. Variability in percent coverover time was evident more in the unman-aged areas than in the NM where percentcover was generally high and invariable(Figure 4). Percent cover was significantlygreater in one or both managed areas thanin the unmanaged areas on seven of theeight sample dates (Figure 4). Cover aver-aged 54.4 percent in managed ponds and12.7 percent in unmanaged ponds.Differences were also apparent betweenthe two managed areas—the NM had ahigher percent cover than the SM on fiveof the eight sample dates (Figure 4).Percent cover in the two unmanaged areasseldom differed from each other and neverwas greater than in the managed areas.

Percent cover and biomass data indi-cated that SAV abundance was greater inmanaged than in unmanaged areas, butpercent cover and biomass were not lin-early related (Figure 5). Percent coverestimates varied with SAV abundanceonly when SAV abundance was relativelylow. Biomass estimates varied with SAV

ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005 239

NM

NUM

SM

SUM

1998 1999 2000

Sep

Jan

May Se

pJa

nM

ay

80

60

40

20

0

Bio

mas

s (g

m-2)

Figure 3. The mean biomass of submergent aquatic vegetation in the four study areas (NM =

North Managed, NUM = North Unmanaged, SM = South Managed, SUM = South Unmanaged)

on Marsh Island, Louisiana. Data is from October 1998 through May 2000 and is presented at

95-percent confidence intervals.

Table 1. Abundance estimates for aquatic vegetation, bareground, and algae inponds sampled using three methods at Marsh Island, Louisiana from 1998 to 2000.n = total number of samples from which means were calculated freq = number ofsamples in which a species was present.

TRANSECT COREn freq Cover (%) SD n freq Biomass(g/m2) SD

Widgeon grass (Ruppia maritima) 192 137 30.5 34Aboveground 324 99 4.9 21Belowground 324 85 2.7 15

Watermilfoil (Myriophyllum spicatum) 192 22 2.3 10 324 2 0.03 0Coontail (Ceratophyllum demersum) 192 15 1.8 8 324 0 0.0 0Pondweed (Potomageton pusillis) 192 1 0.1 1 324 5 0.1 1Spikerush (Eleocharis parvula) 192 11 2.5 12 324 2 0.06 1Bacopa (Bacopa monneri) 192 1 0.2 2 324 0 0.0 0Bareground 192 167 64.7 36Algae 120 47 17.8 33

abundance only when SAV abundancewas relatively high (Figures 3 and 4). As aconsequence, biomass was low and invari-ant as cover ranged from 0 to 50 percent,and percent cover was high and invariantas biomass exceeded 10 g/m2 (Figure 5).

Water and Soil CharacteristicsAfter accounting for the effects of areaand time, we found an inverse relation-

ship between water level and percentcover (Pwater level = 0.0039, r = -0.496), butnot biomass (Pwater level = 0.9375, r = -0.016). That is, as water level increased,SAV percent cover decreased. We alsonoted that water levels changed over timedifferently among the four areas (Parea *time < 0.0001; Table 2). Managementincreased water levels during the winterand reduced water levels during the sum-mer, causing the daily mean water level

relative to marsh surface to range from -2.0 to 2.2 ft (-0.61 to 0.66 m). In March1999, the crest of weirs was 5.9 inches (15cm) below marsh elevation to hold waterin the managed ponds for waterfowl. Thecrests of the weirs were lowered in May1999 to allow water to be drawn off theponds to allow emergent vegetativegrowth. Water level in the SM was mostvariable—being higher than the otherareas in winter and spring, and lower inthe summer (Figure 6).

The SM also had higher soluble reac-tive phosphorus (SRP) concentrationsthan the unmanaged areas on five of theeight sampled dates. The NM also hadhigher concentrations of SRP than theunmanaged areas in September 1999. Wefound no differences in SRP concentra-tions among the areas from October 1999to February 2000. Soluble reactive phos-phorus was more correlated with SAVbiomass than any other water qualitycharacteristic (PSRP = 0.058, r= 0.484;Table 3). As SRP in the water increased,so did SAV biomass.

After accounting for the effects ofarea and time, we found no significantlinear relationship between SAV abun-dance and turbidity, salinity, water tem-perature or DIN (Table 2). Soil mineraland organic content did not differ amongthe four areas (Parea = 0.43). Highest tur-bidity was in late summer whereas lowestturbidity was in midsummer and fall.Southern areas, especially the managedarea, had higher turbidity than the north-ern areas. Turbidity was not correlatedwith SAV abundance (Table 3).

Mean monthly salinity ranged from1.2 to 16.5 parts per thousand (ppt).Salinity in all areas was higher in the sec-ond year of this study (a drought year) thanin the first year, but salinity in the SM wasusually more stable at about 4 to 8 ppt anddid not show the extreme lows and highsshown in the other three areas. The NM,by contrast, usually tracked the unmanagedareas, although it sometimes lagged behindthem during periods of increasing salinityand remained more fresh than unmanagedareas The SM had higher salinity than theother areas in February, March and July1999 and May 2000, but lower salinitythan the other areas in February 2000.

240 ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005

Figure 4. The mean percent cover of submergent aquatic vegetation in the four study areas

(NM = North Managed, NUM = North Unmanaged, SM = South Managed, SUM = South

Unmanaged) on Marsh Island, Louisiana. Data is from October 1998 through May 2000 and is

presented at 95-percent confidence intervals.

NM

NUM

SM

SUM

1998 1999 2000

Sep

Jan

May Se

pJa

nM

ay

120

100

80

60

40

20

0

Perc

ent

Cov

er S

AV

Biomass (g m-2)

100

80

60

40

20

0

Perc

ent

Cov

er

0 10 20 30 40 50 60 70

Figure 5. Percent cover data and biomass data showed no linear relationship. Biomass tended

to be low and invariant when cover ranged from 0 to 50 percent, while percent cover tended

to be high and invariant when biomass exceeded 10 g per m2.

Salinity was not correlated with SAVabundance (Table 3).

Daily mean water temperature rangedfrom 40˚F to 91.8˚F (4.5 to 33.2˚C) andexhibited a typical seasonal pattern. TheSM often had cooler water than the otherareas. Temperature was not correlatedwith SAV abundance (Table 3).

DiscussionWe feel it is important to note that oursites are not atypical. The average biomassof widgeon grass in SAV beds in this study(24 g/m2) was in the low range of previousreports. We could not compare our esti-mates of SAV biomass in ponds (7.6 g/m2)to previous reports because we areunaware of any recorded biomass data ona pond basis. The average percent cover ofwidgeon grass in this study (30 percent)was within the range of other Gulf Coastreports for widgeon grass (Nyman andChabreck 1996, Adair and others 1994).

SAV Survey MethodsOur data suggest that neither biomass norpercent cover fully reflect spatial, tempo-ral, and management-induced variabilityin SAV. We concluded that biomass sam-ples are more suitable than percent coverwhen SAV beds dominate marsh ponds,but percent cover is more suitable whenSAV beds are small and scattered. Ourresults are consistent with those of Adairand others (1994), who concluded thatpercent cover was more sensitive tospecies in low abundance. Percent coverand biomass describe different attributes ofSAV abundance. Percent cover estimatesalone could never reflect all the variabilityin abundance because biomass canincrease after beds cover 100 percent ofthe pond bottom. Theoretically, biomassestimates could reflect all changes inabundance, even when SAV beds weresmall and scattered, if enough cores werecollected from each pond. However, ittook 19 times as long to estimate biomassper pond as it did to estimate percentcover per pond for this study. Althoughsampling five cores per pond was enoughto detect differences among the studyareas, it was not enough to detect differ-

ences when SAV beds were few and small.We calculated that to sufficiently estimateabundance using biomass, would take 105times longer than using percent cover (28cores per pond instead of five per pond).We therefore suggest that it is more effi-cient to estimate abundance with a com-bination of percent cover and biomassestimates than it is to use a small numberof cores only where and when SAV bedsdominate, or to exclusively use a largenumber of cores to estimate biomass.

Other Gulf Coast studies havereported two growing seasons (spring andfall) for widgeon grass (Joanen and Glas-

gow 1965, Pulich, Jr. 1985). Their con-clusion that widgeon grass has two grow-ing seasons has been used as the basis forscheduling SAV sampling (Nyman andChabreck 1996). However, our resultsshow neither clear seasonal patterns norclear winter decline in SAV abundance.We may have failed to find a seasonal pat-tern because of the drought.

Management and Restoration of SAVAlthough water control structures on man-aged study areas were planned, constructed

ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005 241

Table 2. Relationship of the area and time as main effects to water quality charac-teristics, and relationship with submerged aquatic vegetation abundance afteraccounting for any main effects. Data taken from October 1998 to May 2000 atMarsh Island, Louisiana.

Area Time Area and Time Cover BiomassP P P P r P r

Temperature ( ˚C)

and operated together, the results of man-agement were not similar. The NM clearlyhad more SAV than the unmanaged areas.The SM occasionally had more SAV thanthe unmanaged areas, but usually less thanthe north managed area. The generallygreater abundance in SAV in the managedareas indicated that management altered atleast one environmental factor limitingSAV abundance, that being water depth.Water depth was the most likely factoraffecting SAV, but other measured orunmeasured environmental factors alsomay have been important. For instance, apossible reason that we detected effects ofwater depth, but not turbidity or nutrientconcentrations, on SAV abundance is thatwater depth and salinity were measuredhourly throughout most of the 14-monthstudy, but other variables were only mea-sured when we sampled submerged aquaticvegetation. Assuming that the depths ofmanaged ponds were nearly 1 ft (0.3 m)below the depth of other ponds in the area(McGinnis 1997), then water depths lessthan 1 foot but more than 4 inches (0.1 m)were correlated with greatest SAV abun-dance in these highly turbid ponds. Wesuggest that this range may be valuable tomanagers who are attempting to increaseSAV in similar coastal brackish ponds.

The negative relationship betweenwater level and SAV percent cover indi-cates that in managed areas of MarshIsland, alteration of water level, but notsalinity or temperature, affected SAV. Weattribute the negative relationshipbetween water level and SAV percentcover to the attenuation of light in thewater column even though water depth atMarsh Island was much less than otherplaces where widgeon grass has been stud-

ied. In a review of widgeon grass depthlimitations, Kantrud (1991) reported thatthe depth distribution of widgeon grass isgoverned by the susceptibility of bottomsubstrate to wind-induced turbidity. Wesuspect that the bottom substrate atMarsh Island was more susceptible towind-induced turbidity than other placeswhere widgeon grass has been studied.The study ponds were surrounded byLaffitte and Scatlake soils (United StatesDepartment of Agriculture 1979), whichhave 30 to 85 percent clay content(United States Department of Agricul-ture 1984). Those clay contents are muchgreater than in widgeon grass beds inTexas, which range only 0.5 to 21 percentclay (Pulich, Jr. 1985).

In the SM, the marsh was floodedfrom October 1998 to May 1999, whichindicates control structures were inade-quate to drain water. All areas, except theSM, are surrounded by natural ridges ofhigh ground, whereas the SM is sur-rounded by man-made levees and spoilbanks higher than natural ridges, and ahigh, natural beach rim to the south.Water drains to the north because eleva-tions on the island are highest on thesouth and lower toward the north end ofthe island (Orton 1959). Spoil banks andlevees northeast and northwest of the SMrestrict water drainage to the north.Additional water control structures maybe necessary to prevent marsh flooding inthe south managed area. The highest SAVabundance in the SM occurred in the sec-ond year of the study, which was a droughtyear. It may be that the inadequatedrainage was detrimental to SAV growthin a non-drought year and beneficial dur-ing a drought year.

Mean soluble reactive phosphorus(0.1 mg/L) was similar to the mean forChesapeake Bay (0.089 mg/L) where SAVhas been well studied (Boynton and oth-ers 1996). Phosphorus typically decreasesrather than promotes SAV by increasingalgal blooms (Harlin and Thorne-Miller1981), but we found SRP in the watercolumn to be weakly, but positively, cor-related with increased SAV biomass. Forinstance, the SM had the highestrecorded SRP level, but not the highestSAV biomass. For these reasons, we con-clude that phosphorus does not promoteSAV at Marsh Island and the statisticalrelationship was spurious.

SummaryWe found SAV was generally more abun-dant in areas without levees that are man-aged with a water control structure thatallows drawdown and prevents over-marsh water exchange. Although SAVwas generally more abundant in managedthan in unmanaged areas, the effect var-ied among areas and over time. Manage-ment was more effective in increasingSAV abundance in one area than anothereven though both managed areas wereplanned, constructed, and managed in asimilar fashion. The difference in effec-tiveness probably results from differencesin local drainage patterns attributable tospoil banks. Additional water controlstructures may be necessary to preventmarsh flooding in that area. We con-cluded water level governed SAV abun-dance. Submerged aquatic vegetationpercent cover decreased with increasingwater level probably because deeper waterdecreased light availability. The type ofmarsh management that we studied maybe an improvement over traditionalmarsh management techniques because itis less intrusive since soil does not need tobe excavated for levees and less expensivebecause soil does not need to be excavatednor levees constructed. It can, as we havedemonstrated here, effectively increaseSAV, which is an important aspect ofwaterfowl and nekton habitat.

242 ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005

Table 3. Correlation between submerged aquatic vegetation abundance using twosampling methods and water quality characteristics. Data from October 1998 toMay 2000 at Marsh Island, Louisiana. r = Pearson correlation coefficient

Biomass (g/m2) Cover (%)r P n r P n

Temperature ( ˚C) -0.088 0.662 27 -0.153 0.446 27Level (m) -0.009 0.965 27 0.115 0.568 27Salinity (ppt) -0.094 0.643 27 -0.079 0.696 27Dissolved inorganic nitrogen (mg L-1) -0.022 0.408 16 -0.226 0.399 16Soluble reactive phosphorus (mg L-1) 0.484 0.058 16 0.263 0.324 16Turbidity (NTU) -0.030 0.297 14 -0.300 0.297 14

ECOLOGICAL RESTORATION 23:4 ■ DECEMBER 2005 243

ACKNOWLEDGMENTSFunding and logistical support were providedby a contract from the Louisiana Departmentof Wildlife and Fisheries to J.A. Nyman and T.Michot (contract number CFMS #53756), andby the USGS National Wetlands ResearchCenter, University of Louisiana at Lafayette(Department of Biological Sciences andGraduate Student Organization), and UnitedStates Department of Commerce, NationalOceanic and Atmospheric Administration,National Marine Fisheries Service. We wereassisted in data collection and sample process-ing by Edmond Mouton, Garret Thibodeaux,Sergio Merino, Alejandro Arrivillaga, ScottKemmerer, Ron Boustany, Jacoby Carter, LouisMichot, and Megan Jennings. Robert Twilley,Paul Leberg, Beth Vairin, and Darren Johnsonprovided helpful comments to earlier drafts ofthis manuscript. Any use of trade, product, orfirm names is for descriptive purposes only anddoes not imply endorsement by the UnitedStates government.

REFERENCESAdair, S.E., J.L. Moore and C.P. Onuf. 1994.

Distribution and state of submerged vege-tation in estuaries of the upper Texas coast.Wetlands 14:1120-112.

Adams, D.C. and C.D. Anthony. 1996. Usingrandomization techniques to analyze behav-ioral data. Animal Behavior 51:733-738.

Boesch, D.F, M.N. Josselyn, A.J. Mehta, J.T.Morris, W.K. Nuttle, C.A. Simenstad andD.J.P. Swift. 1994. Scientific assessment ofcoastal wetland loss, restoration and man-agement in Louisiana. Journal of CoastalRestoration Special issue No. 20.

Bonham, C.D. 1989. Measurements for terres-trial vegetation. New York: Wiley.

Boynton, W.R., J.M. Barnes, B.J. Weaver, L.L.Magdeburger and P. Sampou. 1996.Sediment-water fluxes and sediment analy-ses in Chesapeake Bay: Tidal fresh PotomacRiver and Maryland main stem. U.S. ArmyCorp of Engineers Report W-96-1.

Causton, D.R. 1988. An introduction to vegeta-tion analysis: Principles, practice and interpre-tation. London: Unwin Hyman Ltd.

Chabreck, R.H. 1994. Marsh management inLouisiana for production of emergent andaquatic plants. Baton Rouge: LouisianaLandowners Association, Inc.

Daubenmire, R. 1968. Plant communities. NewYork: Harper and Row.

Ellison, A.M., M.D. Dertness and T. Miller.1986. Seasonal patterns in the below-

ground biomass of Spartina alterniflora(Gramineae) across a tidal gradient. Journalof Botany 73:1548-1550.

Esslinger, G. and B. Wilson. 2001. NorthAmerican Waterfowl Management Plan,Gulf Coast Joint Venture: Chenier PlainInitiative. Albuquerque, NM: NorthAmerican Waterfowl Management Plan.

Fores-Verdugo, F.J., J.W. Day, Jr., L. Mee and R.Briseno-Duenas. 1988. Phytoplankton pro-duction and seasonal biomass of seagrass,Ruppia maritima L., in a tropical Mexicanlagoon with an ephemeral inlet. Estuaries11:51-56.

Greig-Smith, P. 1964. Quantitative plant ecol-ogy. London: Butterworths.

Harlin, M.M. and B. Thorne-Miller. 1981.Nutrient enrichment of seagrass beds in aRhode Island coastal lagoon. MarineBiology 65:21-229.

Hoese, H.D. and M. Konikoff. 1995. Effects ofmarsh management on fisheries organisms:The compensatory adjustment hypothesis.Estuaries 18:180-197.

Joanen, T. and L.L. Glasgow. 1965. Factorsinfluencing the establishment of wigeon-grass stands in Louisiana. Proceedings of theSoutheastern Association of Game and FishCommission 19:78-92.

Kantrud, H.A. 1991. Wigeongrass (Ruppiamaritima L.): A literature review. Fish andWildlife Research10. Washington, D.C.:U.S. Fish & Wildlife Service.

Kershaw, K.A. and J.H.H. Looney. 1985.Quantitative and dynamic plant ecology.Baltimore, MD: Edward Arnold.

Larrick, W.D. and R.H. Chabreck. 1976.Effects of weirs on aquatic vegetation alongthe Louisiana coast. Thirtieth Annual Con-ference of the Southeastern Association ofGame and Fish Commission 13:100-115.

Louisiana Department of Natural Resources(LDNR). 1994. Monitoring plan (revised)TV-06 Marsh Island control structures.Downloaded 29 July, 2003 fromwww.savelawetlands.org/site/Projects/tv06/tv06.html

Louisiana Wild Life and Fisheries Commission.1959. Biannual Progress Report, 1958-1959, W-29-RII. New Orleans: LouisianaWild Life and Fisheries Commission.

McGinnis, T. E. 1997. Shoreline movementand soil strength in a Louisiana coastalmarsh. MS thesis, University of South-western Louisiana.

Michot, T.C. 1996. Marsh loss in coastalLouisiana: implications for management ofNorth American Anatidae. Gibier FauneSauvage Game and Wildlife 13:941-957.

Moore, P.D. and S.B. Chapman. 1986. Methodsin plant ecology. Boston: Blackwell Scien-tific Publications.

Nyman, J.A. and R.H. Chabreck. 1996. Someeffects of weir management on coastalmarsh aquatic vegetation and implicationsto waterfowl management. Gulf of MexicoScience 14:16-25.

Orton, E.W. 1959. A geological study of MarshIsland, Iberia Parish, Louisiana. Technicalreport of the Louisiana Wild Life and Fish-eries Commission. New Orleans: LouisianaWildlife and Fisheries Commission, RefugeDivision.

Pulich, W.M., Jr. 1985. Seasonal growthdynamics of Ruppia maritima L. s.l. andHalodule wrightii (Aschers) in southernTexas and evaluation of sediment fertilitystatus. Aquatic Botany 23:53-66.

Sokal, R.R. and F.J. Rohlf. 1995. Biometry: Theprinciples and practice of statistics in biologicalresearch. Third edition. New York: W.H.Freeman and Company.

Spiller, S.F. and R.H. Chabreck. 1975. Wildlifepopulations in coastal marshes influencedby weirs. Proceedings of the Annual Con-ference of the Southeastern Association ofGame and Fish Commission 29:518-525.

United States Department of Agriculture.1979. Soil survey of Iberia Parish, Louisi-ana. United States Department of Agricul-ture, Soil Conservation Service.

__. 1984. Soil survey of Lafourche Parish,Louisiana. United States Department ofAgriculture, Soil Conservation Service.

United States Environmental ProtectionAgency. 1979. Methods of chemical analy-sis of water and wastes. USEPA-600/4-79-020. Cincinnati, OH.

Joy H. Merino is a fisheries biologist with theNational Marine Fisheries Service, 646 CajundomeBlvd, Lafayette, LA 70506; 337/291-2109, Fax337/291-2106, joy.merino@noaa.gov

John A. Nyman is an assistant professor with Schoolof Renewable Natural Resources, Louisiana StateUniversity, Baton Rouge, LA 70803-5604; 225/578-4220, Fax 225/578-4227, jnyman@lsu.edu

Thomas Michot is a research wildlife biologist withUSGS National Wetlands Research Center, 700Cajundome Blvd, Lafayette, LA 70506; 337/266-8664, tommy_michot@usgs.gov

http://www.ingentaconnect.com/content/external-references?article=0160-8347()11L.51[aid=6986966]http://www.ingentaconnect.com/content/external-references?article=0160-8347()11L.51[aid=6986966]http://www.ingentaconnect.com/content/external-references?article=0025-3162()65L.21[aid=6986965]http://www.ingentaconnect.com/content/external-references?article=0025-3162()65L.21[aid=6986965]http://www.ingentaconnect.com/content/external-references?article=0304-3770()23L.53[aid=6986963]http://www.savelawetlands.org/site/Projects/tv06/tv06.htmlhttp://www.savelawetlands.org/site/Projects/tv06/tv06.html