COASTAL LAND LOSS AND WKTLANB RESTORATION

tSI R. E. Turner

estuaryare causallyrelated to the landlosses this sealevel ri se,climate change~, soil type,geomorphic century." I then comparethe strengthof this frameworkand age, subsidence or tnanagement. hypothesisto someof theother hypothesized causes of land loss on this coast, There are laboratoryand Four Hypotheses small-scale field trials that support various hypotheses,It seemsto me thatthe mostreliable Four hypothesesabout the causes of indirect interpretationsare basedon what happensin the wetlandlosses in BaratariaBay will be addressed field, andnot on the resultsof computermodels, here adapted from Turner 1997!: laboratorystudies or conceptualdiagrams. H l. i ct n ences of The test results discussed herein are derived t !tin oil banks v d solelyfrom data derived at a landscapescale. The 'ori of 1 loss sin h data set is restricted to a discussion of the Barataria watershed. This watershed is a significant H2. componentof theLouisiana coastal zorie 14,000 lv ha!and there are a varietyof habitatdata available i tl on it. Its easternboundary is the MississippiRiver from whichoccasional overflowing waters are v n.vi hypothesizedto deliver enoughsediinents and on 1 v tno I freshwaterto significantlyinfluence the balanceof rit f i land lossor gain in the receivingwatershed, and whosere-introduction would restore the estuary's wetlands. Improvingour understandingof the H4. w rin si ecologicalprocesses operating in this watershed h ' ' of mightassist in the managementof others.

The effect of geologicalsubsidence and sea DIrect and Indirect Causes of Wetland Loss level rise are not included in this list because both factorshave remained relatively stablethis century Wetlandloss is essentiallythe same as land loss when the land-loss rates rose and fell, Local on thiscoast Baurnann and Turner 1990!. We can subsidencecaused by oil andgas fluid withdrawal discriminatebetween wetland loss that is a direct in Louisianahas been estitnatedto be relatively consequenceof humanactivities, and the lossesthat insignificant comparedto soil sttbsidencerates arean indirect consequence of various other factors. Martin and Serdengecti,1984; Suhayda 1987!. The initial habitat conversions from humati Thereare clearlyIong-term variations in wind, and activities,or "direct impacts", are about 12% of the therefore sea level. However, the trend in water- total land lossesin the Louisianacoastal zone from level rise for the last 80 yearsis essentiallylinear, the1930s to l 990 Britsch and Dunbar 1993!. These and there is no acceleration in relative sea level rise directimpacts are almost exclusively the resultof at any tide gagesite with a long-termrecord up to dtedgingfor oil and gas exploration and recovery, the 1990s Turner 1991!. as well as navigationchannels, Additional direct itnpactsarose from failed agricultural impound- The four hypothesesidentified above were ments.The 'indirect losses' make up the retnaining examinedusing data on a varietyof habitatchanges 88% of all landlosses. Some of thecauses of these obtainedfrom photo-interpretationsof both7 I/2' indirectlosses, or impacts,include reductions in and 15' quadranglemaps that cover the Barataria sedimentsupply, dredging, from subsurface fluid watershed. These analysesinclude documentation withdrawal,or hydrologicalterations. The ratio of of the numberof new pondsforming nearcanals, direct:indirectimpacts resulting frotn human the numberof pondsfilling in near canals.,and the activitiesmay vary under influences such as global amountof landloss and canal density over various BaratariaEstuary Wetiand t-oss 185 time periods. The major question asked is: "Docs land lossresult from the hydrologic changesarising from dredged canals and the consequentialspoil 80 bankparallel to the canal?" 60 Spatial RelationshipsBetween Land i~ and HydrologicChange; 7 l/2' Quadrangle Maps

A newlydredged canal is typically >20 m wide and is 5 rn deepand has a spoil hank built from the 'IG dredgedmaterials that is several meters wide and many times higher than a natural levee. If 0 hydrologicchanges cause wetland loss,then land- 0 3 lossrates should be higher ncarcr,rather than farther, Distance nl from a dredgedcanal and spoil bank. Wetland fragmentationinto ponds is presuinedto bethe first 800;. stage of wetland loss. This hypothesis was Y = 0.<6x+ 14.1 e I examinedusing data on thc spatial distributionof c 6001 k- -0.58 different sized ponds found in 1955/56 and 1978 c fromsixty-three pairs of USFWS 7 1/2'quadrangle e 400' habitatmaps Turner and Rao 1990!. Pondsthat hadmerged or enlargedto becamepart of a larger zimI open-waterbody during the intervalwere identified 0 0 00 400 600 800 lOGO 1 00 and not included in the analysis. The total land loss l978 Canal Area ha1 examinedrepresented 38% of thetotal landloss for the coastalzone in the sameperiod. Someponds Fig. 1. Top:The re!aiionshipbetween the percentof foundin 1955/56were not presentin 1978,and the pondsthat arenew, persisting.and ephemeral for the vast tnajorityof pondswere new. The pondsin interval 1955/6 to 78 ! and distance to the nearesi Baratariawatershed and elsewhere! between I and canal. Bottom: The re}arionshipbetween the areaof 50 ha were the most numerous n = 1104! and tnostly new ponds<60 ha formed between1955/6-78 and canalsurface area ha! in Baraiariaestuary adapted formed after 1955/56 n=935!, from Turner and Rao 1990!. The appearanceof 'new ponds and the persistenceof existingponds waspositively related to the distancefrom the canal Fig. 1, top panel!- regressionof'the two variables pond area and canal Morethan half of all new andpersisting ponds, but area!gave an intercept statistically indistinguishable lessthan 10 % ofthe ephemeral ponds, were within from zero p = 0.03!. Thesepatterns were also 1 km of a canal. If canals had no effect on new documentedin the neighboringSt. Bernardand pondformation, then the distributionpattern of new, Terrebonnewatershecls Turner and Rao 1990!, The persistingand ephemeral ponds should overlap. The hypothesisthat canals and spoil banks caused new frequencydistribution of these pondsdo overlap pondformation is notrejected. withinthree kin of thecanal. The greatestdisparity betweenthe distributionof ephemeraland new TemporalRelationships Between Land Loss pondswas within 1 km of the canal. andHydrologic Change: Salt Marshes

The area of new pondsbetween 0 and 60 ha Thetemporal relationships between canal area that formed between 1955/56 and 1978 in each and landloss for the St. Bernard,Barataria, and quadranglemap was positivelyrelated to the area Terrebonnewatersheds were investigated by Bass «canals in 1978 Fig, 1, bottompanel!. A linear and Turner 997! using aerial photographs.

BaratanaEstuary Wetland Loss lIT

/pat}a} and Temporal ~ 1930s to 195ps;R,- = 0,aq geh}tionsh}psBetween Land 01930s to 1974 R2 0 95 Lossand HydrologicChange: 01930s to 1983; R- =0.93 15' Quadrangle Maps 12000 v1930a to 1990 R2 0 91

Data from 15' quadrangle mapsare available Britsch and 8000 Dunbar 1993! to test for a spatial and temporal relationship betweenlandloss and hydrologic ~g 4MN changes.Britsch and Dunbar's land inventories for coastal 0 Louisiana from the 1930s to 1990 0 500 1000 1500 MOO 500 3000 were derived from colored Direct ha! overlays on 15 base tnaps approximately63 X 10' ha!. They usedthese 0.3 overlaysto map the open-waterhabitat changes betweenmapping intervals. Theseauthors used a consistent photo-interpretationmethod that is slightly differentfrom methodsothers have used, The data set is based on grossland-loss rates, rather thannet land-loss rates, and represents the only data setof its kind for Louisiana! that goes back to the 1930s and that covers the whole coast in a consistent manner.The }nappingdates were from the 1930s 0 0. 0,01 0.02 0,03 0.04 0.05 range 1931 to 1949; p,=1934!, the mid-1950s range Direct Loaa 1951 to 1958; p.=1957!, 1974, 1983 and 1990, Land Britschand Dunbar 993! classified Man-made loss Fig. 4. The relationshipbetween direct land lossand as land that became open water as a direct indirectland loss primarilycanal density!in the consequenceof hurzummodification. Natural loss 8 aratanawatershed for eight 15' quadranglemaps was all other land loss. The 'Man-made loss' in the analyzedby Britsch and Dunbar 993!. No data was Britschand Dunbar 993! analysisis the sameas excluded. Four different mapping intervals are what1 consideredto bedirect land loss in this paper. coinpared: 1930sto 1950s,1930s to 1950s,1930s ro The 15' quadranglemaps included all of the 1974, and 1930sto 1990. Top: Area of direct land loss vs. area of indirect land loss ha!. A linear regression Barataria watershed. of the data is showntogether with the Coefficient of Determination R'!. p < 0.01 in all cases!. Bottom: T~er 997! usedall of thesedata except those The percentindirect land loss vs. the percent indirect niapswith >85% openwater, or <10' ha land within landloss. A polynomial fit of the data is shown a 15'quadrangle map. Only oneof nine mapsthat togetherwith the R'-for eachdata sei p c 0.01 in ail cases!. included the Barataria watershed was excluded in thatanalysis. This one map includedthe Delta Farmsregion, where an agricultu}alimpoundment failedin the1960s and the area became open water. Thisdramatic conversion toopen water represented the Barataria watershed and the indirect land loss Aevast majority of landloss from the 1930sto 1990 Fig. 4; p < 0.01 in all cases!.The interceptv as fortllat quadrangle map. zero,or less,indicating that no significant nct land losswouM occur withoutdredging. identical results There was a strong,positive relationship wereobtained for theneighboring Terrebonneand bet eenthe direct land loss within each 15 mapin St. Bernard watersheds Turner 1997!. 1M R. E. Turner

Thisdata set was also used to plot the relative a low signal-to-noiseratio. ln otherwords, if the changesin land loss since the 1930s. The first competinghypotheses are individually strong or derivativeof thepolynomial regression equation interactwith each other, then a plotof directand describingthe cumulative land loss as a function indirectland losses should look like the scatter-plot of time yielded an instantaneousland-loss rate of datapoints shown in thebottom side of Fig.6, whosetrajectory approached zero in the 1990s, notthe rather strong linear regression shown in the land-lossrates declined as rapidlyas theyrose. top panel, which is the result for the Barataria These estimated instantaneous land-loss rates watershed R' = 0,98;p =0.001!, This result sug- paralleldredging activity. but land losslagged geststhat the cft'ect of canaldredging and hydrologic dredgingactivity by several years Fig, 5!. changeon land loss is quite robust. Testsof CompetingHypotheses HO: Thereare many reasons why it isnot easyto Hydrologic Change developlandscape-scale data andthen usethem fruitfullyto test competing hypotheses about how n 0.3 hydrologicalterations of the landscapeaffect the N 0 area ol coastal wetlands. Underneaththe surface 7 0.2 of theLouisiana coast are deltas of differentages V iti andcomposition. Theamplitude and energy of the 0.j. tides within and amongwatersheds is not C homogenous.Thedensity and timing of dredging activity variesamong watcrsheds because of economicand anciently-defined geological factors. 0 025 0.05 Furthermore.ifthe four hypotheses identified earher Direct Loss in thispaper are compiemcntary, or if the causal Land significantterms interact, then the effectsof dredgingshould be difficult to tease out because of HO: Salinity, Suspended Sediments and Flood e mrect t and ass ' total Landlossr10 o Permitted Direct t andloss!

~ 8 M O X

0.025 0.05 Direct Loss l 960 >980 year Land

Fig 5 L d lossand cmg d Kgng over t me for the Fig.6, TopPaneh the hypothesizcdrelationshipbetween Louisianacoastal zone from Tutner 1997!. Permitted indirectand direct loss if thehydrologic change area open squares! isbased on the area dredged each hypothesisiscorrect. Bottom Panel: the hypothesized yearthat is permiued bythe State's coastal zone relationship between indirect and direct loss if the managementprogram. The other two estimates are hydrologicchange hypothesis isnot correct or if several basedona statisticalfii of the Brit sch and Dunbar hypothesisare significant individually or actingin l993>land loss data for the whole coast. concert.The left panel from Fig. 4 in Turner1997! is the actualdata for the Baratariawatershed. BaratariaEstuary Wetland Loss 189

Another test of competing hypotheses is to the reducedsediment loading. There are at least comparethe relationship between land-loss rates and three linkedassumptions if this hypothesisis true: hydrologic changein the Baratariawatershed for I! saltwater changesmust occur,throughout the theyears before and after the significantdrop in watershed,and not just in onesmall part of it, ! suspendedload occurred in thc mid-1950s. If this thesechanges must be significant plant stressors and sediment-loaddecline was an additional cause of lead to vegetationlosses, and, ! otheremergent wetlandloss, then theintercept of theland-loss and plantcommunities do not successfullyre-colonize canal-densityrcgrcss ion line shouldbe higherafter any newniche created. the 1950s, han for data for beforethe 1950s Fig, 7!. If the decline in sediment loading were an The 'saltwater intrusion' and the 'sediment insignificantinfiuence on land loss during the deprivation'hypotheses were tested by assigning mappinginterval!, then there would be no difference each 15' basemap a numberto denoteits relative in theslope of the regressionlines for the two data position inland from the estuarine entrance sets.The hypothesized influcnced was not observed describedin Turner 1997!. The quadranglemap see Fig, 4!, so we shouJdreject this competing closest to the estuarine entrance was the lowest hypothesis. numberedmap, and the quadrang Je inap furthermost from the seawas nurnbercdthe highest. The relative effects of increased estuarine salinity which are tLttlgenerally observed on this Thesetwo hvpotheses salinity stressand coast;Wiseman et al., 1990! and flood protection sedimentdeprivation! can be rejectedfor several andnavigation levees on land-lossrates can also be reasons,in additionto the previously described testedwith the data setof Britsch and Dunbar 993; relationships.The estuarineheadwaters are where Turner1997!. The hypothesissuggests that land- overbankflooding previouslyoccurred but not Joss rates will increase if freshwater fiow into everyyear!. Plant stressresulting from sediment- estuarineheadwaters is restrictedand plants become deprivationor salt water intrusionshould be: I ! stressedby an increasein salinity,or, becauseof lowestfor theplants that are inost adaptedto high salinity and nearest the reinaining suspended sedimentsupply the estuarinecrttrance!, and, 2! Ho: Hydroloqic Change highest for thoseplants least adapted to salt, i.e., thoseplants located in the freshwaterrnarshes. 1f and Suspended Sediments salinity stressoccurs, and if it leads to land loss. After then the amountof indirect land loss per direct land 1955 loss shouldtheoretically have been highest in the tn 0.3 lowest salimty zone bottom panel in Fig. 8!. This Ol ef ore result was not observed, The indirect:direct land 955 loss ratio declines, not increases, going from salt to V c 0.2 freshwaterzones along a gradientfrom the estuarine mouth to inland. This ratio does not decline in the regionwhere plant stressesshould bemost sensiti ve t 0.1 to salinitystress Fig. 9; R' = 0.80;p =O,G] 6!, There is moreland lossper dredgedchannel in the high- 0.025 0.05 salinityzone of thc estuarythan in the low-salinity zone, and land loss with dredging where flood Direct Loss protectionlevee s aremost likely to reduceoverbank Land flooding. Although saltwater intrusion may be significantlocally, it doesnot appearto be a major Fig, 7. The hypothesizedrelationship between indirect anddirect land lossfor two mappmgintervals if indirect factor driving land loss this century in this suspen~ sedimentloading in the l950s hasan effect basin,These two hypothesesare therefore rejected. on landloss. This patternwas not observed see Fig. 4l. Ho: Hydrologic Change 5 hig O V 4 OV 3 5 7 Distance{relative! seaward! landward!

low Fig.9. Therelationship between the land loss per w Distance W areacanal Y axis! andthe distanceto the coastin Seaward Landward 15' quadranglemaps. Land loss is from Britschand Dunbar993! and is for the 1930sto 1990. The distancemeasure is a simplemap sequence from coast to inland,

Coatelusiosts HO Increased Salinity or Sediment Oeprivation Thchypothesis that hydrologic change is the hig primarycause of wetlandlosses in the Barataria watershedwas tested in variousways using data on V landscape-'scalehabitatchanges. Comparati ve tests ofcompeting hypotheses were also made to exainine the relativestrength of theinfluence of salinity O8! V changes,suspended sediment load reduction, and flood protectionlevees on land-lossrates. The resultswere consistent among several tests. The formationofponds is highestnear canals, whereas pondsthat revert to wetlandsare relative!y scarce t low nearcanals. When canal density is high,land-loss ~ Distance ~ ratesare high, and when canal dredging slows, then Seewnrd Landward land-lossrates stabilize or decrease.Thc amount of landloss per area of dredgedcanal is highest wherethe hydrologic gradient is highest from Fig,8. Thehypothesized re!ationship between the tides!,and not in the estuarineheadwaters. The indireiclossrates Y axis! vs. direct losses Xaxis! hydrologicchange hypothesis isthe only acceptable fromthe 1930s to 1990 for each estuary asa function hypothesisofthe four that were tested. lt explains of distancefrom the coast data are for the ] 5' therise and fall of' land-loss rates in time and space quadranglemaps examined byBritsch and Dunhar l993!.Only quadrangle mapswith <85% open water in the Baratariawatershed and is a moreefficient and>8 ! 00ha are included. The top panel indicates the explanationthan alternative hypotheses t"e resultsnecessary tosupport thehydro!ogle change applicationof Occam'sRazor!. hypothesis.andthe bottom panel isthe anticipated resultif increased salinity, decreased suspended Theseconclusions obviously h»e som sedimentsorboth is significant. Theactual resuhs are thatshown in the top pane! and Fig. 9. consequencesforconserving theexisting wet!~+ and for their rehabilitationand restoration. BaratanaEstuary Wetland Loss %91 consequenceis that thc role of plantsin wetland Universitycourse -Wetland Loss. Restoration and stability should be consideredas more than a Managetncnt"are especially thanked for their reactionto patterns in inorganic sediment rates. patienceand wiHingness toleam, sharc, and critique planbare major contributors to wetland accretion, thevarious ideas that survived here, and for helpin plant ecologists could contribute to wetland buryingothers, restorationby investigating belowgroundplant organicstorage pools; wetlandhydrologic studies LIT ERAT Li RE C1TED should be expanded to include landscape-scale interactions. We have much to learn. BAss,A. AhoR. E. TuRRRR.1997. Relationships betweensalt marshloss and dredged canals in Landscape-scaleanalyses bring a uniqueset of three southLouisiana estuaries. Journal r>f complicationsto ecosystemscience. Obviously Coastal Research 13:895-903 landscapesare not homogenous, and therefore BAUMANN,R, H, AND R. E. TDRhnR. ]990. Direct variability is introduced when comparing one impactsof outer continentalshelf activities on watershedto another. Further, varying the size of wetland loss in the central Gulf of Mexico. the measuringunit tnay confound detectionof the EnvironmentalGertlogy and 8'ater Resources actualpattern. An analogymight be the exatnpleof 15:189-198. tryingto determineif theearth i.scurved. Although BOESCtl,D. F., M. N. JOSSFJYh, A. J. MeirrA, J. T. the earth's surface is easily seen as curved when Moaxis, W. K. Nt,TTLF., C. A. SIMnh'sTAD,AVt! flying long distances, its shapedoes not appearto D, J, P. Swum. 1994. Scientific Assessment ol be round when walking, One cannot quantitatively Coastal Wetland Loss, Restoration and detect this curvature with ordinary surveying Managementin Louisiana.Journal of Coastal instrumentsas easilyalong a 10 m path, as alonga ResearchSpecial Issue. No. 20. 10 km path. If a 100 km measuringpath or 100 BRrrscii, L. DAND J, B, Duh'uAR, 1993. Land loss km'plot! hasa mountainrange, then aneven longer rates: Louisiana Coastal Plain. Journal of path may be neededto compensatefor the local Coastal Research 9:324-338. variation introducedby geological history. The MARrtv, J. C. ANDS. SHtnelsoacn. 1984. Subsidence relationshipbetween hydrologic change and land overoil andgas fields. Reviews in Engitreering lossmay not seem evident by analyzinga 1-hapixel, Geology6:23-34. insteadof a 100-ha pixel. or if the wetland is ScAtFE,W. W., R. F. TDRNeR,Ahio R. COSTAvzx. encompassedby stranded beach ridges, or 1983. Recent land loss and canal impacts in embeddedwithin the Mississippi River delta, or coastal Louisiana, EnvirorrmentalAfanagetnent isolatedby remnantchannel leveesburied a few 7:433-442. meters below the surface. This situation leads one SuHAYDA,J. N. 19g7. Subsidenccand sea level. to understandthat the interrelationships between Chapter10, 1n: R. E.Turner and D, R. Cahoon wetlandhydrology and plant health may extend over eds,!: Causesof WetlandLoss in the Coastal kilometers,and that evaluatingthese connections Central Gulf of Mexico. Voh 2: Technical mayinvolve some ratherinnovative and humbling Narrative.Final reportsubmitted to Minerals failuresbefore we fully appreciatethe proper scales ManagementService, Contract No. 14-12-001- necessaryto investigateand quantitatively predict 30252. OCS Study/MMS 87-0119, New theconsequences of coastal wet.land management. Orleans, LA. TuRhrER,R. E, 1990. Landscapedevelopment and ACKNOWLEDG MENTS coastalwetland losses in the northernGulf of Mexico, American Zoologist 30:89-105. Theseanalyses accumulated over tnany years, TL~zR, R. E. 1991. Tidegage records, water lc vcl andespecially through the often vigorous discussion rise and subsidencein the northern Gulf of withcolleagues and the hardwork of the variousco- Mexico. Estuaries 14:139-147. authorsand studentswho sharedthe researchefforts TtnthrER,R. E. 1997. Wetlandloss in the northern escribedherein. Students of the Louisiana State Gulf of Mexico; Multiple working hypotheses. Estuaries 20:1-13. 192 R. E. Turner

Tva~ R. E. ~n Y.S. Rao. 1990.Relationships hetweenwetland fragmentationand recent hydrologicchanges in a deltaiccoast. Estuaries 13:272-28 ] . WtSmae, W. J., JR.,E. M. SwENSOt',mn J. PowER. 1990, Salinitytrends in Louisianaestuaries. Estuaries 13:265-271. Pattern and Process of Land Loss in the Louisiana Coastal Zone: An Analysis of Spatial and Temporal Patterns of Wetland Habitat Change' JOHNW. DAv,JR.', GAttv P. SHAFFEtt', LoUls D. BRITSCH-', DFN1SEJ. REED, SUZANNF. R. HAWES, DONALD CAHOON"

'Depanrnentof Oceanographyand Coastal Sciences and Coastal Ecology Institute, Loui.sianaState University,Baton Rouge,LA 7N303, TEL: 225-388-6508;email: johnday Otlsu. edu 'Departmentof BiologicalSciences, Southeastern Louisiana University Hammond, LA 70402-0736;TEL 504-549-2865; email: shafe@selu,edu 'U.S.Army Corps of Engineers,New Orleans DistrictEngineering Division P.O, Box 60267, New Orleans, IA 70I60-0267; TEL: 504-862-2995;email: louis.d bri tschC+mvn02.usace. army,mil 'Departmentof Geology,University of NewOrleans, IVew Orleans, LA 70148; TEL: 504-280-7395; emai L dj [email protected] 'U.S,Army Corpsof Engineers,New Orleans DistrictEngineering Divisiort P.O. Box 60267, New Orlearts, LA 70I60-0267; TEI.: 504-862-25I8; email: suzanne.r.hawes@'mvn02. usace.army. mi I National WetlandsResearch Center, U.S. GeologicalSurvey, 700Caj undome Blvd., Lafayette 70506; TEL: 3I8-266-8634; emaiL. don cahoon@usgs,gov

ABSTRACT:An earlierinvestigation Turner 1997! concluded that auostof thecoastal wetland loss in Louisianawas caused by theeffects of canaldredging, that loss was near zero in theahsence of canttlsvand that land loss hect decrettsed to nearzero by the late 1990's. This analysh was based on a 15-minutequadrangle approximately 68,000 ha! scalethat is toolarge to isolateprocesses responsiblefor small-scalewetland loss and too small to caphtrethose responsible for large-scale hrss.Herein, we conduct a furtherevaluation of therelationship between direct loss due to canal drcdghtgand all other loss From 1933-1990 using a spatialscale of 4,100ha that accurately captures localland-loss processes. Data sets for the Pontchartrain and Breton basins did not meet the criteria for parametricstatistics. Regressions of other wetlami loss on canalarea i~., directloss! for the Blrdfoot,Terrebone, and Calcasleubasins were not signlHcant. Significant positive curvilinear relationshipswere isolated for the Baratarla RMAO! and Mertnentau R0.29! basins,indicating that tbeextent oF canals is signi6caatly related to wetbmd loss in these basins. A significantnegative relationship R~~l! wasfottnd for theAtchafalaya coastal basin which had statisticaHy lower loss rates than the otherbasins as a whole. Whenthe data werecombined for aH basins,92% of the variation in other wetland loss was attributabieto cana!s. AHsignificant regressions intercepted the Y-axisat potdtiveloss values indicating that smne loss occurred m theabsence of canals. We agree with Tttrnerthat canalsare an importantagent in caushtgwetland loss in coastalLouisiana, but stronglydisagree that they ase responsible for the vast majority of this loss. We conclude that wetland lossin the Missimippidelta is an ongoingcomplex process invoiviag several interacting factors and that efforts to create and esture Louisiana'scoastal wetlands ntust emphasizeriverine inputs of Fresh water and sedbnents.

Fromthe SymposiumRecent Research in CoastatLouisiana: 'This paper ia s condensed version of an article thol can tttaturatSystem Function ansi Response ta Humantnfluence. be Foundin Vohsme23 of Estuariestutd is printed here Rozas,L.P., !.A. Nymao,C.E. Pro%tt,V.N. Rabalais,DJ. with the permiaaiooof the Estuarine Research Reed,attd R,E. Tumor editors!. 1999. Publishedhy Loutstaoa Federtstion. SeaGrattt Co/lege Program.

193 104 J,W. Day et at.

Iatrodtsetlots 3. Land I ossis highestnear the coast and decreases Turner997! analyzedwetland loss in coastal inland. This implies that land loss is highest in Louisianawetlands from 1932to 1990 using salt marsh and lowest in fresh marsh and that statisticalanalyses of landto waterchanges in 15- sal waterintrusion has not been an important minute quadranglemaps. He concluded that factor in land loss. virtuallyall of the losswas causedby canals,that land loss was near zero in the absenceof canals, 4, Restrictedriverine input has had little impacton andthat land loss rates declined to nearzero by thc landloss. A corollaryto this i» that land lossin late 1990s, Turner explicitly excluded saltwater wetlandsin the AtchafalayaDelta region,which intrusionand the leveeingof the MississippiRiver is not leveedand receivesapproxiinately one as important factorscontributing to land Joss thirdof the totaJAow, of the Mississippi River, becauseland losswas highest near the coastand has been the saine as that of other basins with becauseregressions of direct versusindirect land similardensity of canals. If land loss in the losshad zero intercepts. In thispaper and in Turner Atchafalayadelta region is low, it wouldsu ggest 997!, direct land loss refers to wetland which that~bile lack of riverineinput may not have becomeswater when canals are dredged and indirect directly caused land loss, riverine input is land loss refers to all other wetland loss. Turner' s essentialto building newwetlands and reducing conclusionscontradict PenJand et al. 996!, who lossof existingwetlands, concluded that about 46% of wetland Jossin coastaJ Louisianahas occurred through natural processes. Materials and Methods

The objectiveof thispaper is to catTyout a Data Base further evaluation of the causesof land loss in the Louisianacoastal zone, particularly with respect to 8ritsch andDunbar 1993!quantified wetland the efFectsof canals, A more detailed version of to water changesduring four snappingintervals: thispaper Day et al. 2000!may be found in the early-1930sto rnid-1950s referredto in this paper journalEstuaries. This reduced version has been as 1932-1955;the actualdates of thedifferent snaps printedwith permissionof Estuaries, usedvaried slightly!, 1956-1973, 1974-1982,and 1983-1990.They aggregated their direct and indirect Turner997! statedfour hypotheses about the landloss data to the standard15-minute quadrangle coastwide causesof landloss and testedthem with map scale -68,000 ha or 168,000 acres!, This is a variousstatistical analyses of directand indirect Jand convenientscale for mapsbecause it is largeenough Jossrates from 15-minute quadrangle snaps which to presenta considerableamount of detail but small containabout 68,000 ha. Wehave incorporated the enough so that the number of maps which essenceof thesein thefollowing hypotheses which encompassthe entire coastalzone about50 maps! we testin thispaper. is notexcessive, Britsch and Dunbaralso produced six spatial maps that were color coded for each Hypokhetses mappinginterval Britschand Dunbar 1996,Fig 2 of Day et al. 2000! 1. Directland loss i.e., dueto canals!is quan- titativelyrelated to landloss in general,both for However,the 15-nunutequadrangle scale is individualhydrologic basins and for thcentire generallytoo large for statisticalanalysis of site- coast.indicating that most land loss can be specific patterns of wetland loss in coastal attributed to canals. Louisiana.The patch sizeof mostof the land loss that hasoccurred in the coastalzone is considerably 2. Whenshrect land lossis zero,other land lossis smallerthan the scaleof a 1 5-minutequadrangle. closeto zero i.e.,the intercept in regressions of An exatninationof the mapsof Britschand Dunbar otherland losson directland lossis zero!. 993! shows that most 15-minute quadrangles RelaionshlpBetween Canals and Wetland}l QM $95 containseparate patches of wetland lossthat arenot wetland habitat type in each cell, we overlaid spatiallyor functionallyrelated to eachother see vegetationc}assification maps on the habitatlos» Fig,2of Day etal. 20 N!. For example,in adetailed nMps. For the 1932-1955loss period, we usedthc studyof the proximityof canalsand wetland land 1949 mapof O' Neil 949!. In a similar manner, loss Leibowitz 1989!,no relationshipexisted for for the 1955-1973, 1974-1982, and 1983-1990 }os }andlos» greater than 5 kin froin the canals. intervals,we used vegetation classification map» for 1968 Chabrecket. al, 1'968!, 1978 Chabreckand Data Analysis Linscombe 1978!, and l988 Chabreck and Linscombe 1988!, respectively. To better approximatethe locationand scale of manyof the land loss processesand patterns Regression and ANOXIA analyses w ere including oil and gas fields where most canals performedusing the SYSTAT 6,1 general linear occur!, we subsampledthe tnapsof Britsch and rrxx}e}.Specific hypothesis tests between direct }and Dunbar 996! at a 4,}00 ha 0,100 acres! or 6,4 loss i.e., due to canals! and other land lossi nc1ude: by 6.4km ce}l size, which approximates the size of ! otherland loss in the entire coastalzone and in an average oil field network or a large marsh eachhydmlogic basin is statistically relatedto direct managementimpoundnient Cahoon and Groat land lossand ! when direct land loss is zero, other 1990!, A priori, the total number of cells n=121! land loss is zem; ! land loss is highest near the wasdetermined by budget constraints,These cells coastand in salt rnarshes for this test,distance from werearranged along 19 transectspositioned nearly the coastwas defined by the sequenceof each cell equidistantfrom eachother, such that the number in the subsamplingtransects!; and ! land loss in of cellsper unit area of thecoast was roughly equal. theAtchafalaya Delta hydrologic unitis not different We did not include transects which would have been froin that in other hydrologic units. HypothesesI practicallyall water eg transectsin LakeCalcasieu and2 weretested with regressioninode}s, wherea.s or Lake Pontchatrain!.However, to ensurethat each 3 and4 were testedas separateone-way ANOVAs basinreceived at least n=7 cells, it was necessaryto usingbasin, cell number in thetransects, and marsh place some transectscloser together e.g., the typeas categoricalvariab}es. birdfoothydrologic unit is verynarrow compared to the others!. The transects were positioned We did not eliminate cells containingless than perpendicularto the localcoastline either the 6u}f 15% lard in 1933 as in Turner997! becausethis shoreline or interior bay margins as in the potentiallycreates two biases. First, by definition, Terrebonne,Barataria, Breton, and Pontchartrain these cells are located on shorelines where the basins!and variedin lengthbecause they extended priinarycause of landloss is waveerosion, Pen}and to theinland extent of the mappedwetland, bui did et al, 996! reportedthat about35% of landlos» notoverlap one another.While this processwas wasdue to wave erosion Second, very few oil field not completely random, the placetnentof the canalnetworks arc located directly on the Louisiana transectswas inadc without. regard to overa}lland coastlineand henseelimination of coastal cel}» may losspatterns and, once the coastalcell of each improvethe regressionfit, transectwas p}aced, a}l ccl}s along the transectwere fixedin place. Log transformationsof the dependentor independentvariable, or both, wererequired in In eachcell, direct dueto canaldredging! and several instances when the data did not meet the all other ]andloss termednatural loss by Britsch criteriafor parainetricanalysis and when relation- and Dunbar 1996! was determined for each shipswere curvilinear. Nonparametric regression samplinginterval. SinceBritsch and Dunbaronly wasrequired for the Pontchartrain andBreton Sound classifiedchanges from land to water,the change datasets Burkesand Dodge 1993! Whendata frommarsh to spoilbank adjacem to eachcanal is contained an observation with excessive le'veiage not counted as loss. Britsch and Dunbar did not i.e., > 0.55!the results are presented with and classifyland loss by habitattype. Todetermine the withoutthat observation Hair et a}.1998!. For ihe 18' J.W, Oey at al.

Barataria,Atchafalaya, and combined data set, one for the Birdfoot, Terrebonne, and Calcasieu observationrequired omission because it produced hydrologicunits bestfits producingp>0,12, Fig. an excessive >+3.0! Studentized residual Hair ei 2!. Positiverelationships existed for thc Breton al. 1998!. Unlessspecified, differences were Sound F,,=10.37,p=0.023, Fig. 3a!, Barataria deemedstatistically significant at alpha= 0.05. F, =56.97,p<0.0001, Fig, 3b!. andMerinentau An attempt was made to analyze the data basins F~~ 1073 p 0004 Fig. 3c!, producing standardizedto the amountof landpresent at each R -0.675,R'=0.721, and R'=0.349, respectively. site in 1932 as in Turner 1997!. That is, the direct For the Baratariadata set, omission of thcsingle andother loss values were propottianalized io the observationonthe Y axis with leverage&.57,Fig, amountof hmdin eachtx:11 at the beginningof the 6aj decreased the R' from 0.721 to 0.470. study.These data required an arcsinesquare root Interestittgly,in the Atchafalaya,which is a basin transformationtomeet norma!i ty andhomogeneity characterizedby high input of riverineinorganic of varianceassumptions Sokal andRohlf 1995!, sedimettts,a significant negative relationship existed

4000[ A I Rggggitt!lt:For all the data standardizedto the amountof land presentin 1932, no significant Blrdfoot relationshipsoccurred, regardless of transforination Fig, 1!. interestingly,when cells were reinoved 2 3000 thatcontained less than 15% land in 1932 seeFig, 4 of Turner. 1997, for comparison!,the fit did p becometnarginally significant F, rr3 4.16, p<0.044!albeit weak RM.036!. Forthe analyses $ 2000- by basin,standardization did not itnprove the fits; therefore,all subsequent interpretations are hmited to the non-standardized data.

tDOO Statisticallysignificant relationships between 0 400 200 300 other land lossand direct land losswere not obtained Direct Loss ha! 9- 06 B. Terrebonrt e sI- rrr 0.5

p 7" 4- rn zs 03|

Ca p E 0.2~. P ~ J . 5- z 0 14 4'" 0~ 0 1 00 200 300 400 0.0 0.01 0.02 0.03 0 04 0.05 Direct Loss ha! Normalized Direct Loss Fig. 2. Direct land !oss and other land loss for the A-! Fig. l. Directwetlands loss and otherwetlands loss Birdfoot and B.! Terrebonehydrologic unib from forall basins normalized tothe land present in each 1933-1990. Data for Calcasieu Basin are not shown as quadratin 1933 as in Fig.4 ofTurner 1997!. n = 121. directloss was zero for all quadrats. RehionshipBetween Canalsand WetlandLoss t tlat between other wetland loss and direct loss cti F W,93, p=0.041,Fig. 3d!, producingRa=0,236. 3 7 Forthe three parametric fits Barataria,Merrncntua, andAtchafalaya!, the intercept b ! wasstatistically g 56 greaterthan zero, indicating that there was wetland 4 loss in the absence of direct losses.

The regressionof otherwetland loss on direct wetland loss with all of the data combined could 0300 1 2 23456 78 accountfor only 9.29oof wetland loss attributable tocanals F, = .00,p=0.001, R'=0.092, Fig. 4!. Rank of Direct Loss Theintercept of b.=644ha was highly significantly 9 differentfrom zero t=6.60, p.0001!. 87 >~VI.: The data set with all basins combined was also used to test for diffetences in h 6 other wetland loss from 1932 to 1990 due to effects is! 5 of basins,the distance from the coast, and marsh type Fig. 5!. The basineffect washighly significant for theraw F, ,=9.56, pc0.0001,Fig, 5a! and F40 12345 678 standardized F, o,=g.71,pc0,000! data and thc a prr'orilinear contrast of theAtchafa! aya versus other Log Direct Loss ha! 8 ~ ~ basinswas highly significant F~nz 12 9 Cl pc0.0001and F, , =11.37,p<0.0001 for the raw C, and standardizeddata, respectively!, despite relativelylow lossrates in the BretonSound and Pontchartrainbasins Fig. 5a!. Land lossdid not differ from the coast inland for the raw data 5 F=l.32, p&.247!, butdid for the norlnalized O 4 ' Mermerttau data F,,~=2.68, pW.014! with the highest losses occurringin bothcoastal and the mostinland cells F 0 1 234567 Fig.5b!, Wet!andloss did not differ for different Log Direct Loss ha! marshtypes F,, =2,72,p=0,069!, but displayed a similarpattern as that of distancefrom the coast D. Fig. 5c!. 8 6 Discussion I Theresults of ouranalysis indicate that canals ~ ~ w ~ havebeen an importantfactor in landloss, but that 0 therelationship with land loss varies between basins. F Atchafalaya ForBarataria Basin, up to 72% of otherwetland 4 2 3 4 5 6 losswas statistically associated with direct losses dueto canals. However, omission of a singleouther Log DirectLoss ha! decreasedthe R' to0.47, or by 35%. Frolnboth an ecologica!and statistical perspective, webelieve that Fig. 3, Regressionsof other wetlands loss on direct a singleobservation with thatmuch leverage war- wetlandsloss for A,! BretonSound. B.l Baratana" C.! Mermentau,and D.! Aichafalaya basinsfrom rantsomission. There wasno significantrelation- 1933-1990 95% confidence interval s of the mean are shipbetween direct land lossand other land lossfor included. foe J,W. payat al.

4000~-r r r All Sites Combined

Q 0 300

Ci ~ 0 8 ~ 2 M N ,, 0

Fig. 4. Regressionof otherwetland loss on direct c, C3 wetland loss lor «ll hydrologicunits combinedfor penod af 1933-1990.95% confidencelimits of the 5 meafl are Included,

E 01 several of the basins Pontchartrain,Birdfoot, Terrebonne,and Calcasieu!. This does not mean, of course,that land loss has not been caused by i 0 1 2 3 4 5 6 7 canalsin thesebasins, Rather, it indicatesthat other processesarc also i tnportant and inaskthe potential &stance from Coast cell ntimter! statisticalrelationship, orthat insu%cient statistical 15 powerexisted due to low sample sizes. There was no clear patterrtin landloss with distance from the

coastor fordifferent types of marsh, and in general 10 theAtchafalaya hydrologic unit, with the highest C! D Q! riverineinput. has had a significantlylower rate of l4 wetland lossthan the other basins. io E05 Althoughthe t5-minutequadrangle is a convenientmapping scale for a varietyof purposes, i t commingles i.e.,aliases! the functional processes responsiblefor coastal wetland loss in Louisiana. 00 In a detailedstudy of theproximiry of canalsto patchesof high wetland land loss, Leibowitz I989! ~G concludedthat canals were not the prirrtary cause Marsh fype ofthe loss. In addition, when positive relationships betweencanals and land lossexisted, the Fig, 5. Differencesin wetlandloss rates from 1933-1990 relationshipdecomposed within 5 kmof thecanal. among A! basins, B! disumccfrom the coast Cell 1 is Our quadrNswere 6.4 krn ona sideand therefore neamstthe coast!. and C! marshtype. Standarderror shouldhave captured positive relationships crisply, barsare included.For basins,BA is Barataria,BS is had they existed. BretonSound, CS is Calcasieu,lvlo is MississippiDelta, ME is Mennentau, PO is Pontchartrain, TE is Terrebonne,and TV is Atchafalaya ReiaionsbipBetween Canals and WetlandLass 199

We found that direct land loss that is, dredged patternof landloss acrossthe salinity gradient, canals!ranged from beingnot statisticallyrelated There are documented cases, however, where wc to other land loss in several basins to accountirtg believesaltwater intrusion was responsible for !and for 47% and 68% of the variation in other land loss loss.For example, huge marsh losses occurred west in the Barataria and Breton Sound basins, respec- of the northernpart of Lake Calcasieuduring the tive!y.Based on regressionana!ysis, Turner ! 997! 1956-1973mapping interval, These losses occurred conc!udedthat canals arc responsiblefor the afterthe completion of theCa!casieu Ship Channe! majorityof wetlandloss and that whendirect !oss in 1941 and fol!owed the passageof Hurricane is zero,other land lossis zero. Our analysis detnon- Audreyin 1957. The hurricaneapparently led to stratesthat this is not the case. The scale of the 15- massivesaltwater intrusion and widespreaddeath minute quadrangles used by Turner leads to of the heshwaterCladium marsheswhich previously statisticalre!ationships betv ccn canals and other occupiedthe area Morganet a]. 1958!. The land !osswhich is clearly unrc!atedto cana!s i.e., remainingmarshes are now intertnediate tobrackish. shorelinewave erosion!. Pcn!and et a!. 996! also The constructionof the MississippiRiver Gu!f concludedthat land loss was due to multiple Outlet !ed to sa!twater intrusion and caused the death interactive causes and indicated that about ha]f of of almost all of the Taxodium swamps which the land loss was natura! and that thc total amount formerlyoccurred east of the MississippiRiver of! and loss indirectly due to "submergence"because below New Orleans Coastal Environments 1972!. uf human modifications was about 35%, Among Someof this area is now openwater, but much of the causes of land loss other than canals were wave the swamp has converted to Spartina marsh erosionand tectonic fauhing. scatteredwith ghostcypress trunks,

In general,we found a loweroverall statistical Turner 997! conc!udesthat the isolationof re!ationship between direct and other land loss than mostof thedeltaic plain from riverine input by f!ood is suggestedby theresults of Pen!andet al. 996!. control!evees has not played a significant role in Forthe combineddata sct. the R'-was on]y 0.092, wetlandloss. This contradictsa long historyof Onereason for this is that the re!ationshipwithin research that demonstrates how the river built and eachbasin varies from linearto log-linear to log- maintainsthe delta Fisk et al. 1954; Kolb and Van !og to no re!ationshipto a negativerelationship in Lopik1958; Day et al. 1995,1997; Roberts 1997 I. the Atchafalaya Delta area. The mostobvious example of this are the marshes in the Atchafalayadelta region, a non-]evecdcoastal An importantdetrimental impact of canalspoil bay wheresubaerial land buildinghas continued banksis thatthey lead to the reductionof sediment since 1973 Roberts et al. ]980!. Our analysis inputand poor drainage of marshsoi!s Reed 1992; demonstrates that land loss in this area is sig- Boumansand Day 1994;Cahoon 1994; Cahoon et nificant]ylower thanthat of anyof the othercoastal a]. 1995b; Reed et a!, 1997; Swenson and Turrter basins. !987!.The pmgressive waterlogging due to reduced sedimentinput can interact with existing salinity to Turnersuggests that there is a "needfor tnuch producedeleterious effects on vegetation.A number greaterecological understanding" of wetlandsand of studieshave shown that multiple stresses,such thatthere is litt]eappnxiation of therole p]ants play as salinity and waterlogging,have a much more in "dotninating the accumulation of sediments detrimentalimpact on coasta]vegetation than a throughtheir contributionto soil organicmatter sing]estressor Grime 1979; Mendelssohnand be]ow ground."We certainly agreeon the needfor McKee 1988; Mc Kee and Mendelssohn 1989; Grace further study on the functioning of these systems. and Ti!man 1990; Shaffer et a!. ]992]. But it haslong beenrecognized that soil formation and accretion in much of the Mississippi deltaic We foundno c!earpattern in land lossrates plainis dominatedby organicsoil formationfrom with distance from the coast, or across different rootproduction and managetnent suggesuons have tnarsh types, suggestmgthat there is no general explicitlyincorporated this function Hatton et a!. 200 J.W. Oay et al,

1983;Temp]et and Meyer-Arendt 1988; Cahoon restoration,specifically with reference tocana] spoil 1994;Cahoon et al. ]995a;Day and Temp]et 1989; banks,should be a necessarycomponent of a Dayet al. 1995,1997; Nyman et al, 1993a,b!. holistic,integrated delta restoration p]an. ~s a]one Organicsoil formation often accounts for70%-80% however,will haveminimal impact if it is not of accretion,but additionof mineralsediments coupledwith reintroduction of river water.Mincra] resu]tsinstimulation ofp]ant production and health. sedimentswill begenerally necessary to re>ui]dand Mineralsediment addition, especially in riverwater, maintainthe coast,It is notlikely thatremoval of has several positive impacts on marsh plant spoilbanks will resultin revegetationona large communities.The rrunera]sediments add strength scale.Several studies have shown that there ts rapid «ndbulk to the sediments,for example,and they lossof elevationof 10-15 cm whenp]ant death carrynutrients that stimulate productivity and iron occurs Nyman et al, 1993a,b;Delaune et a], 1994. whichcomplexes with sulfidephytotoxins. Fresh Kernpet al. ]999!. In addition,soil strengthin water also reducessalinity stress. highly stressed,low elevation wetlands is very low andthere is verylittle elevation gain even with high Basedon his analysis,Turner 997! con- accretionrates Cahoon et al. 1995h;Kcrnp et a]. cluded that land loss rates in the coastal zone will 1999!. A nutnberof actionscan help reduceland be closeto zero by theyear 2000, But land loss loss,but only riverine input can leadta major ratescontinue to bc high,with measuredloss rates creation of new land. This agreeswith our rangingfrom 65 to 9] km'yr' in the 1980sand the fundamentalunderstanding of how deltas function 1990s Barras et al, ]994!, ln addition,landscape bothspatially and tempora]ly Roberts 1997; Day modelingresults Reyes et al. ]999! andstatistica] et al. 1997!. If Louisianais ever to achieveno net pmjectionsbased an past land loss rates J. Suhayda, loss of its coastal wetlands, we believe that it wi]] Dept.Civil Engineering,LSU, personalcom- be necessaryto expandrestoration strategies to munication! indicate that ]and loss rates will includemajor river diversions. continueto behigh aver the coming decades. For his modeling,Turner used a second-degree ACKNOWLEDGMENTS polynomialregression to fit wetlandloss over time andto extrapolateloss rates from ]991-1995. This Wethank Bruce Pugesek and Darren Johnson typeof regression canfit only a symmetricparabola, for statisticalcomments, and Donald Boesch, James andextrapolation with polynomial regression in Gosselink,Martha Hixon, Andrew Nyman and four generalis cautioned Zar ]996!, anonytnousreviewers for critiquingthe manuscript. Supportfor this effortwas partia]]y provided by the Conclusions LouisianaSea Grant College Program and by the U.S.Artny Corps of Engineers. We concludethat wetland loss in the MississippiDelta is a verycomplex process and that LITKRATURE CITED lassis causedbya dynamicand interacting setof processes.We agreewith Turnerthat. canals have B~ J A p E BouaGIKNsAND L. R. HA~~ '. been,and continue to be,an important agent in 1994 Land loss tn coastal Lou>siana 1956 contributingtothis land lass, We disagree, however, 1990, NationalBiological Survey, National thatcanals are responsible fora majority ofthe land WetlandsResearch Center Open File Repn lossand that land loss iszero when canal density is 94-0] 4 pp 10color prints zero.The exclusion of sediments, freshwater, and BocMAvs,R. M, ANDJ,W. DAv. 1994,Effects « nutrientsof the Mississippi River from much of the coastalzone has elinunated a major land building twoLouisiana marsh management plans on andmaintenance mechanism which historica]l water and material f]ux and short ter> counteractedmany of the processes responsible for sedimentation. Werlands14: 247-261- landloss and thus is a major factor in coastal land BRrrscH,L D, ANDJ, B Dtns~AR 1993.Land lo lossinLouisiana. Wea]so agree that hydrologic rates:Louisiana coastal plain. Journal af Coasral Research 9: 324-338, RelaionsiIIpBetween Canals and Wetland Loss 201

D. AND J, B, DL'NBAR, 1996. Land loss DAY,J,W., D, PONT,P.HENSEL. ANo c. IBAr'Fz.1995. in coastal Louisiana. A series of seven color Itnpactsof scalevel rise on deltasin thc Gulf mapsof coastalland loss. Tech, Rept, GL-90- ofMexico and the Mediterranean.. Thei tnpor- 2, Maps 1-7. U.S. Army EngineerWaterways tanceof pulsingevents to sustainahility. ExperimentStation, Vicksburg, MS andU,S, Estuaries18: 636-647. ArmyEngineer District. New Orleans, LA. DAY,J,W. JRJ.F, MARTIN, L. CARooclt.ANn P. H. BIIRKES,D. ANOY- DODGE-1'993, Alternative TEMPLET.1997. System functioning asa hasis Methodsof Regression.Wiley, New York. 228 for sustainabletnanagetnent of deltaic pp. ecosystenLsCoastal ljrfana gement 25: 11S-153. CANOON,D, 1994. Recent accretion in twOmanaged DAY,J,W. JRG,P. sNAF»st, I..D. BRITscH,D.J. Rluux tnarshimpoundments in coastal Louisiana. S,R, HAwEs,ANn D. CAHOON,2000. Pattcm EcologicalApplications 4: 166-176. andprocess of land loss in the Louisiana CANooN,D, R., ANDC. G. GRoA7,EDs., 1990. A coastalzone: an analysis of spatial and studyof marshtnanagement practice in coastal tempondpatterns of wetlandhabitat change. Louisiana, 3 vols, Final report submitted to Estuaries23: in press. MineralsManagetnent Service, New Orleans, DELAum, R, DJ. A. NYMANAND W. H. PATRlcK, Louisiana. Contract No. 14-12-0001-30410. JR. 1994.Peat collapse, ponding and wetland OCSStudy/MMS 90-0075. lossin a rapidly submergingcoastal marsh. CAnooN, DD.J. REED, Avo J.W, DAY, 1995a, Journalof CoastalResearch 10, pp.1021- Estimatingshallow subsidence in microtidal 1030. salt tnarshes of the southeastern United States: Ftsx, H, N., E. MCFARLAvJR., C.R. KoLe, ANoJ. L, Kaye and Barghoorn revisited. Marine WILBFsrrJR. 1954. Seditnentaryframework Geology128: 1-9. of the modernMississippi delta, Journal of CAIIooN, D., D. REEO, J.W. DAY, G, STEYER,R. Sedirnentary Petrology 24: 76-99. BOUMANS,J. LYNCH, D. MCNALLYANo N. LATIF. GRACE,J.B. ANoD. TILMAv,1990. Perspectives on 1995b, The influence of Hurricane Andrew on Plant Competition.Academic Press, San sediment distribution in Louisiana coastal Diego,CA. 484 pp, tnarshes,Journal of Coastal Research 18: GRnvtE,J, P. 1979. PlantStrategies and Vegetatio~ 280-294. Processes.John Wiley, New York.222 pp. CHABRECK,R. H. ANo G. LINSCOMBE. 1978. HAIR, J. F., R, E, ANOERsoN,R. L. TATHAM,Avn W. Vegetative-Type Map of theLouisiana Coastal C, BLAcK.1998. MultivariateData Analysis Marshes.Louisiana Department of Wildlife with Readings,Fifth Ed. Macmillan,New and Fisheries, New Orleans, LA, York, Inc. 745 pp. CHABRECX,R. H. ANO G. LINscOMBE. 1988, HATtoN,R., R. DELAONE,ANo w. PATRICK.1983. VegetativeType Map of theLouisiana Coastal Sedimentation,accretion, and subsidence in Marshes.Louieiana Department Of WIIditfe tnarshes of Barataria Basin, Louisiana. Lim- and Fisheries, New Orleans, LA. nologyand Oceanography 28: 494-S02. CHABRECK,R, H., T. JOAv~, AlsoA. W. PALMISANo KEMPP., J.w. DAY, D.J. Rrro, D.R, CAHOON,Axo M. 1968. Vegetative Type Map of Louisiana WANo.1999. Predictingeffect.s of sea level CoastalMarshes. Louisiana Department of riseand sediment supply on surfaceelevation Wildlifeand Fisheries, New Orleans. in anorganic rich salt marsh. In: L.P.Rozas, COASTALENVIRoNMFNTs, INc. 1972. EnvirOnmen4d J.A.Nytnan, C.E, Proffitt, N.N. Rabalais,D.J. BaselineStudy, St. Bernard Parish, Louisiana. Reed,and R.E. Turner cds.!, Proceedings, Preparedfor St,Bernard LA ParishPolice Jury. Recent Research in Coastal Louisiana. 159pp. Unpublished. LouisianaSea Grant College Program, LS1;, DAY,J,W. AND p. TEMPLET1989, Consequences of BatonRouge, LA, sea level rise: Implications from the KoLa,C. R. AvDJ. R. vANLot tx. 1958.Geology MississippiDelta. Coastal Management 17: of theMississippi Dehaic Plain -Southeastern 241-257, Louisiana. US Army EngineersWaterways 202 J,W.Day et al.

ExperimentStation. Vicksburg, MS. Technical ReED, D. J., N. DF. LucA ANo A. L Poove. 1997. Report 2: 3-482. Effect of hydrologic managementon marsh LElaowrrz,S.G. 1989.The patternand processof surface sediment deposition in coastal land tossin coastalLouisiana: A landscape Louisiana. Estuaries 20: 301-311. ecologicalapproach. Ph.D. Dissertation, REvesE., J.F. ManrtN, M,L. Wutrr:, J.W. Dav, G 3'. LouisianaState University, Baton Rouge, LA, KEMF,In Press.Landscape modeling of coastal 298 pp, habitat change in the Mississippi delta. MCKRR,K. ANDI, MEND~SOxN.1989. Responae Of Ecology. a freshwatermarsh plant communityto in- RORRRrs,H. H. 1997, Dynamic changesof thC creasedsalinity and increasedwater level, holoceneMississippi River delta plain: the AquaticBotarry 34: 30]-316. delta cycle. Journal of Coastalkesearch 13: MRNDELssxoN,I. AND K. McKER.1988, Spartr'na 605-627, alter@iflora die-backin Louisiana:time-course RORRRVS,H. H, R, D. AD*ass, axo R. H. W, investigationof soil waterloggingeffects, CuvxtNoxAM. l980. Evolution of sand- Journalof Ecology76: 509-521. donunatedsubaerial phase, Atchafalaya Delta. MOROax,J, P., L. G, XtcxoLs, ANDle. WatOHr. 1958. Louisiana. The American Association of Morphologicaleffects of HurricaneAudrey on PetroleumGeologist Bulletin 64: 264-279. theLouir iana coast. Technical Report No, 10, SttARTR,G, P, C. E. SasSER,J. G, GOSSELtgtt,aNn ContributionNo, 58-3, BatonRouge, LA: M. REtMANEK.1992. Vegetationdynamics in LouisianaState University,Coastal Studies the emergentAtchafalaya Delta, Louisiana, Institute,53 pp. USA. Journal of Ecology80: 677-687. NVSraN,J.A., J. C, CALLAWAY,axn R, D, DzLatjm.. SoxaL,R, ANDF. ROLxF.1995 BiOmetry,Third 1993a.Case Study of a rapidlysubmerging Edition. W.H. Freeman and Co., New York, coastalenvironment: relationships among 887 pp, verticalaccretion, carbon cycling, and marsh SwENsON,E. ANDR. TuRNzR.1987, SpOil banks' less in Terrebonne Basin, Louisiana, Pro- effectson a coastaltnarsh water level regime. ceedingsof the Hilton Head Island South Estuarine, Coastal, and Shelf Science24: 599- CarolinaU.S.A. InternationalSymposium 2: 609. 452-457. T~ruw, P. ANDK. MFYFRARENDT. 1988, ~iaiana NYMaN,J,AR.D. DFLauNR,H.H. Roantrrs,axo wetlandloss: a regionalwater management W,H. ParRtcx,JR. 19931, RelatiOnship approachto the problem. Environmental betweenvegetation and soil formationin a Management 12!: 181-192. rapidlysubmerging coastal marsh. Marine TuRNFz, R.E. 1997. Wetland lOSsin the nOrthern Gulf FcologvProgress Series 96: 269-279. of Mexico: multiple working hypothesis. O'Natl, T, 1949. The muskratin the Louisiana Estuaries 20: 1-13. coastalmarshes with map of theSouthern Part ZAR, J. H, 1996, BiostatisticalAnalysis, Third of Louisianashowing vegetation types. Edition. PrenticeHall. Upper SaddleRiver, Louisiana Wild Life and Fisheries NJ. 662 pp. Commission.,Technical Rept. 28 pp. PFNLAND,S., J. MFNDHsSOttN, L. WaVNF.,aND D, BRrrscx.1996. Natural and Human Causes of Coastal Land Loss in Louisiana, Coastal StudiesInstitute and Wetland Biogeochemistry Institute,Louisiana State University, Baton Rouge,LA, 25 pp, R~>n, D. J, 1992. Effect of weirs on sediment depositionin Louisiana coastalmarshes. Envi ronrtrenta Management 16: 55-65.

2N E, Reyes et al.

INTRODUCTION affectingit changethroughout time Boumansand Sklar1990!. This ypeof feedbackhas been tested Coastal wetlands across southeastern inaquatic modeling programs like LAPTER Reyes Louisianahave cotnpacted and are being lost as they et al. 1994!, and used in terrestrial simulation convertedto open water Wells 1996!. Annual prograinsPATCHMOD Wu andLevin 1994!and variability in mean sea level MSL! can be several ECOLECON Liu et al. 19941. centimetersper year Baumann1980!, and can result in in~ penetrationof salinityinto wetlands STUDY AREA Penlandet al, 1988!.Such changes are believed to underliethe generalpattern of displaceinentof The Barataria-Terrcbonncsystcin is an freshwatervegetation by tnoresalinity tolerant ittterdistributaryestuarine-wetland system located communities,and vegetationdie-off followed by in southernLouisiana. The Baratariabasin portion conversion to apen water Roberts 1997; Wells is located between the natural levees of the 1996!, MississippiRiver and BayouLafourche. It is roughlytriangular in shapewith an areaof 6100 Most wetlands in the Barataria - Terrebonne kin'. TheTerrebonne basiri is borderedby Bayou estuarinecomplex are losingelevation to MSL at Lafourcheon the east and the Atchafalaya River on variablerates Penlandand Ramsey1990!. Reed the westand occupies 5500 km' Fig. 1!, 995! calculated mean annual land loss rates in Barataria at 20.1 km' for 1958 to 1978 and 34.5 Bothbasins are dynainic systems undergoing kmi for 1978 to 1988, and for Terrebonneat 24.9 changedue to naturaland human processes, The and 31.6 km', respectively.Reed 995! also Baratariabasin hns been closed to direct ri ver inflow estimated that indirect land loss due to canal since1904. Precipitation provides its mainsource dredgingcould accountfor tnore than 30% in of freshwater;however, the Mississippi River exerts Baiatariaand 10% in Tetrebonne,It appears that an indirectinfluence on salinityin the lowerbasin local anthropogenic modifications have had by reducingsalinity in thenearshore Gulf of Mexico differenteffects in each basin and making it difficult Petretet al. 197I !. The Terrebannebasin is directly to assesschanges on a regionalbasis. influencedby the AtchafalayaRiver. As a resuh, the westernportion of thisbasin is one of thcfcw Understandingthese habitat changes iscritical locationsin southernLouisiana that has experienced to assessthe long-termeffects of proposed net landgain Boeschet al. 1994!. restorationapproaches, The objectives of thisstudy wereio ! constructa multiplescale process model The vegetativecornrnunities in bath basins for the Baratariaand Terrebonnewatersheds to reflect gradients in elevation and supply ot understandand predict regional habitat change and freshwater.Marshes occur as bandsof sah, brackish. ! assesslong-term indirect and cumulative impacts andfresh vegetation from theGulf inland Chabreck of managementalternatives asproposed for wetland and Condtey1979!. Fresh marshesgive way to restoration. swainpsand bottomland hardwoods at higher elevations,These communities have overlapping Wchave been addressing these issues using salinity tolerances when grown in the laboratory. spatially articulatedlandscape models Costanza et butgenerally competition leads to distinctzonauon al. 1990;Sklar et al. 1985;White 1991; White et al. in the natural setting Conner et al. 1987!. 1991!.These dynamic spatial interaction models havebeen called Coastal Ecological Landscape MKTHGDS SpatialSimulation CELSS! by Sklar and Costanza 991!. CELSS modelsincorporate location- A landscapehabitat prediction model was»dt specificalgorithms that allow feedback between the for eachbasin. Each is a dynamicspatial model localprocesses andthe landscape dynamics, sothat usingvariable time and spatial scales. Both models boththe landscape andthe intensity of the processes usca finitedifference, 2-dimensional and vertically Landscape4lodeiing in Coastei Louisiana 204

100 ktu

scale

Atchafal ayaRtv

~ upland

1 BayouPerot Grand IsLe

»g.1, The State of Louisianashowing location for the Batataria and Teirebonne watersheds. Nutnbcrs indicate time senesslations.

»tegratedhydrologic modulewith a time step dt! modulewhich redefines the habitatinosaic cell size of 1hour and a spatialcell sizeof 100km' coupled of I km'-!on a bi-annualbasis. The conceptual witha primaryproductivity module with a dtof ] frameworkof the four modulesis presentedin day and 1 km-'cell size. Output from the Figure2. The model was written in FORTRANand hydrodynamicand productivitytnodules are runson a UMX platfortnand specific details for subtnittedto a soilmodule dt of I yrcell sizeof 1 the modulesinteraction can be found elsewhere ktn'! andthen evaluated by a habita.tswitching Reyeset al in review.;White et al. 1997}. 20S E. Reyeset at

Fig.2, Flow of calculations forthe landscape progrsni modules indicating time and spatial scales,

Hydrodynatnics velocity,is determinedto depositon the marsh or resuspended,Deposition is calculatedas a netdaily The hydrodynamicmodule uses the diffusion value,and salinity and duration of floodingare wave approximationfor shallow waterto calculate averagedto daily values. watermovement and sediment transport Singh and Aravamuthanl 995!. This approximation requires Productivity thailocal acceleration, uniform flow, and Coriolis forceio be considered neghgible. This simplification The macrophytemodule was viewed as a is necessarybecause standard hydrodynamic representationof any portion of a givenlandscape equationsrequire smal ler time steps than is practical with homogenouscharacteristics. The biological for longterm predictions Cheng et al. 1993!.The moduleruns on a daily time-step,integrating nei effectof frictionis accounted for by a Manning productivity for the inacrophyte community coefficientthat is the 100 km' average ofal l I km' throughouta I km' cell. Althoughstructurally the habitatdependent Manning coefficients included in same, the biological inodule uses different that100 km' cell. Inputs include rainfall, pumping parametersand initial values for eachhabitat once outfall,and rivers. Outputs are limited toevaporation is spatiallydistributed, Specifically, two biological andtidal boundaryexchanges. Infiltration into components are modeled: non-photosynthetic groundwateris assumed negligible. carbon biomass roots and stems! and photosynthetic carbon bioinass leaves!. Gross Initialspatial inputs include a 100kin' land pmductionis a functionof biomass,maximum gross elevationmap and a suspendedsediinents field. productionrate and a limitationfunction Hopkinson Resultanthourly values for 100 km' water elevation, et al. 1988!. This limitation function includes watervelocity, salinity and suspended sediment responsesto waterlevel. salinityand temperature. distributionsarelinearly interpolated toyield 1 km-' The waterloggingfunction represents the different values.interpolated water and landelevations are tolerances of each habitat as determined by a usedto calculate water depth. Suspended sediment representative specie! to flooding conditions is evaluatedhourly, and according to the water Nyrnan et al. 1993!, Salinity stress is also LandscapeModeling in CoastalLouisiana 206 determinedby habitat type Pezeshkiet al. 1987!, Thc habitatswitcher module has two corn- The temperature response function varies ponents:a dailycounter and a switcher.Thc daily seasonally.Respiration rates proxy for metabolic counterqueries biomass density, sal inity and depth stresses!adjust to thesesame factors Burdick et al. of standingwater at each1 km-'cell. Basedon these 1989;Dai andWieg crt 1996;Hopkinson et al. 1978!, values, habitat type is determined according to Excessfixed carbon is translocated to the non- classificationcriteria Table 1!, and the counter photosyntheticstorage Howes et al. 1985!.The advancesby oneunit. At regularpredeterinined oppositeprocess occurs if respirationlosses offset intervals, the habitat counters for each cell are photosyntheticintake, submittedto the habitatswitcher component Th» algorithmevaluates if a habitatchange shouldtake Soil place under several conditionalrules. If more than half of thetime open-water conditions existed then The soil module includes three components: the cell type is assigrcdas open water Nyman et inorganicsediments, dead belowgroundbiomass al. 1993!.If not, the vegetativehabitat type with andtotal beiow groundorganic sediments dead the highestcount is assigned.When the habitat type biomassplus non-photosyntheticbiomass from the changes,new productivityand respiration rates biologicalmodule!, Both inorganic and organic appropriatefor thathabitat type are applied in the componentswere divided by a constantbulk density biologicalmodule. The habitat switching decisions to calculate elevation. aremade once every two yearsin this model version but any time interval could be defined. Total seditnent elevation was calculated by addingthe heights of inorganicand organic storage Table 1. Habitat typedefinition by salinity PSU! and accounting for percentage pore space. and biomassdensity kg OM m *!. incorporation of aboveground litter to the belowgroundstorage is assumed negligible Nyrnan et al. 1993!.The amountof inorganicsediments is conservedand a decompositionrate is appliedto belowgroundorganic sediments.

Subsidence accounts for 85-90% of the relative sealevel rise RSLR;estimated at L2 cmfyr!within Louisianacoastal marshes Day and Ternplet 1989!. While subsidenceis notexplicitly includedwithin thesoil module,decomposition of theorganic stock partially simulatesshallow subsidence.Deep subsidenceof the Holocenelayer is accountedfor FprchttgFunctions by increasedrates of eustaticsea-level rise Fenland andRamsey 1990!. Forcingfunctions for the landscapemodel wereinput as time series, The time series necessary Habitat Switcher to run both models from 1956 to 1992 werc cotnpiledfrom 15 stationsfor 9 parameters wind This modulekeeps track of habitatcharac- speedand direction, inorganic suspended sediments, teristicsand environmental parameters for eachcell evaporation,rainfall, air temperature, salinity, tides, throughouttime Theseenvironmental parameters river discharge! aresummarized and evaluatedannually to determine if theyare representative of another habitat type. Daily tide stageswere obtained from thc The interactionof primary productivityand National OceanService NOS! at Bayou Rigard. environme.ntalconditions defines habitat type for 1955-1979and from EastPoint, Grand isle from Laurenrothetal, 1993!.Once a thresholdofchange 1980-1988.Daily salinity values at GrandTerre Lab isreached, the model simulates habitat change. collectedby Louisiana Wildlife and Fisheries were gtP E. Reyes et al.

usedas the boundarycondition for the Barataria Salinity was calibrated by adjusting boundary basin.Boundary conditions for Terrebonnewere set conditions i.e., salinity initial conditions and usingthe salinity distribution reported by Murray diffusionrates!. Salinity results matched closely the andDonley 994!. availabledata Murray and Donlcy 1994! in the lower portionsof the basins. SpatialData Assembly Theinacrophyte production module was tested Habitatmaps for theBarataria and Terrebonne usingspecie specific parameters for eachwetland basinswere provided by USFWSin rasterized625 dominantcommunity type. Varying physiological m' georeferencedmaps. These digital maps were parameterssuch as, tlooding and salinity tolerance derivedfrom aerial photography for 1956and 1978 within reportedranges Howes ct al. 1986;Nyman and satellite imagery for 1988 and 1990, and et al. 1993;Pezeshki et «I. 1987;Visscr et al. 1996; «ggregated to I kin'. In additionto openwater and Wisemanet al, 1990!produced seasonal trends of developedupland, each cell was classifiedas a productivityand bioinass Conner and Day 1976; wetlandtype characterizedby a singledotninant Kirby andGosselink 1976!. species.Salt rnarshes were characterized by Spartina alrerniflora,brackish rnarshes by Spartinapatens, Spatial Calibration fresh marshesinclude Pariicsrnrheiniromon and swampsby Taxodirrm disrichiurri Chabreck 1972; The model was run for the 1978 to 1987 Conneret al. 1987;Tiner ! 993;Visser et al. 1996!. decadeusing the USFWS maps to set initial conditions and final spatial comparison.The CA LIBRATION objectives were: I! to siinulated ecological processesacross the landscape and ! to verifythat Recognizingthat a landscape model is a all the landloss processes were implicitin the complexsystem with non-linearresponses to landscapemodel. This second objective was stated different environmentalstimuli, a calibration to assurerhat although when the landscapemodel strategywas implemented to considerthe effects of doesnot explicitlyincorporate local processes e,g, differentscales and impactsof each moduleon canalsand spoilbanks!,it incorporateswhatever modelperformance, This strategy allowed us to regionaleffects these impacts might have.This increasethe model complexity as each module was calibration method thus compensatesfor local I irsttested individually and then combined into the effectsby loweringthe regionalresilience of the landscapemodel. The data set used for this habitats to land loss. calibrationonly included the parameters from 1978 to 1988, The landscapecalibration required a matchin habitat distribution. This implied consistencyamong TemporalCalibration the land use changes, land loss rate and habitat responseto climatic variability. Previoushabitat Resultingtime seriesfrom each module were modelingefforts in southernLouisiana Sklar and comparedagainst available data at specific Cos anza 1991! have utilized a goodness of fii locations.The model was run repeatedly until these parameter Ft; Costanza 1989! to evaluate the model time series matched the field data. The performance.This fit index comparestwo maps hydrodynamicinodule was tested using the 1994 usinga movingwindow technique and calculates elevationinap Alawady and Al-Taha 1996!, no the degree of accuracy between them in values windsand theoretical tides for 1987 and 1988. rangingfrom 0 no match!to 1.00 perfectrnatch!. Hourlywater levels were computed by the model For the calibration purposes, the model was at cellscorresponding to the locations of theNOS repeatedlyrun comparingresulting mapsfrom 10- dataand compared for consistencyand match. yearsimulations ending in 1988to theUSFWS 1988 Manningcoefficients were then adjusted for each map. This was doneuntil the overall fit improved of thehabitat types to produce the best match. to 0.85 for both basins. an~scapeModeling in CoastaiLouisiana

lnitia] conditions salinity, elevation, and canbe answered using this cotnparatlvc approach. Manningcoefficients!, and biological parameters Whatwere the effects ofthc management alterna- such as salinity tolera.ncesfor each habitat were tives.'r WasW it positiveor negative with rcgardito manipulatedduring the calibration.The 1978- ] 988 landloss' Was there any associated change inhabitat calibratedsimulations yielded a fit of 0,89 for distribution? Baratariaand 0,85 for Terreborme.There was also agreementfor total wetland and water areas for the Normal ConditionsScenario two watersheds Ft = 0.96 for Barataria and Ft = 0 94for Terrebonne!. Table 2 presentsthe resulting TheNC scenarioconsisted of a 30-year fits andhabitat areas for eachof theruns, starting simulation for eachbasin. To run simulationsinto with the calibration,validation, and management the future,theoretical time series and boundary optionexperiments that will be discussedlater in conditionsneeded to bedecided. We repeatedthe ihe text. originaltime seriesin reverseorder, because climate tendsto be cyclic Thomson1995!, The forcing Anothercalibration was doneby comparing functionsand boundary conditions are actual data available land loss information. Annual land-loss for years1955-]992 butwhen thc year 1993 was estimatesfrom 1955/56, 1978 and 1990generated simulated, the climate from 1991 wai used, 1994 from the USFWS and USACOE Dunbar et al. simulationused climate from 1990, and so on. 1992!intervals are given in Table3 alongwith those createdby the model, TheresuJting habitat distribution for Barataria Fig. 3! converted1,105 km-' to openwater during Validation 1988 to 2018 Table2!, The largestdecline 98 km'! wasfor brackishmarsh, while only 5 km-' of To validatethe model, parameters settings for swamp were lost. The modeI iden t ified large the base case 978-1988! were used to simulate portions of the middle and lower brackish marsh 1956-1988conditions using the USFWS map as that convertedto open water, whereas the upper initial conditions.The model predictedland loss basin, dominated by swamp habitat, remained fluctuations for 1956-78 between 0 and 65 km' for relatively unchanged. Barataria and 85 km'-for Terrebonne, similar to reportedvalues of' 73 km'- Gagliano et al. 1981!. Thebrackish marsh sites of BayouPcrot and Habitatand goodness of fit resultsare presented in BayouL'Ours were subjected to time seriesanalyiis Table 2. Goodness of fit values for these simulations Fig. 4!, asthese sites underwent a changefrom were lower than the calibration values. The decrease brackishinarsh to open water. F igure 4a revealsthai. of accuracyin model predictionscan be attributed following 2005, the water level mcreasedand to the differentland Josstrends Table3! through remainedhigh as plant biomass decreased until the the30-years tested. habitat at BayouL'Ours changed in the year2017. At BayouPcrot Fig. 4b!, this habitatswitch MANAGEMENT SCENARIOS occurredin 2015probably because of risingsea level and increased salinity. Landscapemodels of this typeare one of the fewtools that canbe used to predictthe effects of ln Terrebonne, 1,204 km' of land werc complexspatial interactions and cuinulativc, long- convertedto openwater of which660 km- were terrneffects of global changes,Simulations, from freshmarsh! habitat Fig.5!, Thisoccurred moitly 1988to 2018,were performed for a seriesof in a largearea of freshinarsh at the northwestern inanagernentscenarios Table 2!, The first scenario, portiorrof thebasin, although some fragmentation referredto asNortnal Conditions NC!, simulated a of brackishand salt marshesalso occurredin the future continuation of current trends, Later southeast.These changes resulted on a water/land simulationsare evaluated by comparingresults ratio increase from 0.62 to 1.51 by 2018. againstthis NC scenario.The followingquestions 209 E. Reyes et al.

Table2. Susnmaryof scenario results performed iu eachbasin.

ResultingHabitat Coverage ktn'! Calibration Fit

Fresh Swamp Brackish Salt Open Scenario Name Marsh Marsh Marsh Water Land/Water Habitat

Terrebonne Basin 1170 432 828 576 2106 USWFS

Base Case 1100 516 865 551 2080 0.9433 0.8508 calibration; 1977-87!

Base Case 847 657 596 842 2170 0,8664 0.7274 validation; l 956-87!

Normal Conditions 510 428 365 3310 988-201 8!

Freshwater Addition 51] 428 362 3305 0.9829

Hydraulic Restoration 552 429 466 365 3300 0.9829

Barataria Basin 755 734 2952 USWFSmap

Base Case 723 1002 722 634 2854 0.9597 0.8932 calibration; 1977-87!

Base Case 1191 577 2883084 0.8648 0.7439 validation; 1956-87!

Normal Conditions 396 1017 236 217 4057 988-201 8!

FreshwaterAddition 521 1019 311 236 4057 0,9751

StoppingSaltwater 303 1015 226 247 4132 0.9645 Intrusion

Note: Fit valuesfor BaseC asescenarios were computed against 1988 USWFS habitat map, Fit valuesues for ~age~nt opt ons were compute against the2018 no~~ conditions habitat map

Additionof FreshwaterScenario inflow was simulatedat 301.4 m'/s I0,000 c fromDecember through March and 28.3 rn-'/s I,M Thisscenario simulated diversion of river cfs! for the remainderof the year.A hydraulrc waterinto theBarataria basin. The site for this connectionalready exists between Barata freshwaterdiversion was Davis Pond, an area that Terrebonneso the simulatedfreshwater addition encompasses6.6 km' of eastcentral Barataria. implementedinto Terrebonne by increastng Lanttscapa 4lodeling irt Coastal Louis taros 210 fable3. Annual loss rates km'! for the 1956-78 and 197848 periods inyerrebonne andBarat na Basins

Source: Dunbar Reed US FWS Model Dunbar Reed USFWS Mode1 et al. 1992! 1995! output etal, 1992! 1995! oulput

Note: '1958-74 and 'l 983-1990 intervals.

2018 Normal 'ondi tlons 2018 Added fieshsrater

ArealEvent tktnli Patrertt Hahitat T!pe h;ortnat Fresh»ater trash nt~b 393 53 l ssramp l017 l019 hrackish marsh 336 3ll salt marsh 3 lt} 339 open water 40s9 3a33 ttplands l 7ll l 7' ttrnposid i

«g 3.Resulting habitat distribution of Barataria Added Freshwater scenario compared ared vs. Normal Condntnns scenario 211 E. Reyes et al.

a! BayouL'Ours

1.4 lp 1 2 1.0 0,8 0.6 0.4 0,2 0,0 1990 1995 2000 2005 2010 2015 1990 1995 7000 7005 2010 2015

Year Year b! BayouPerot

10 14 1.2 1.0 0.8 o 06 04 02 0.0 1990 1995 2000 2005 2010 2015 1990 1995 7000 2005 2010 2015

Year Year Waar Depth Biomass Fig.4, Time series of waterdepth and photosynthetic biomass for a>Bayou L'Ours and b! Bayou Perot satnple locations for Nortnal Conditions scenario.

ratesthere by 84.9 m'ls ,000 cfs! duringpeak internalwater flow by redirectingflow ftotrtthe west discharge. portionof the basinfarther into the east.For this simulation,Manning coefficients were increased Results ftom this scenarioshowed that land locallyas a proxyfor the modiT~cations{Fig. 6a! lossin Barataria was reduced by l 13 ktrr'compared We assumedthat increasingwater friction in the to NC Fig, 3!. The preservationof freshtnarsh was selectedarea would result in increasedflooding and largely responsiblefor this difference. Habitat redirectflow towardsthe internalportions of the distribution effects were minor in Terrebonne and basin. resulted in almost no difference from the base case scenario Table 2!. The resulting habitat distribution in Terrebonne showed about 42 km' tnore- of fresh Restorationof tuternalHydrants ~nario marshand 33 km' lessof brackishmarsh compared to the NC Fig. 5!, This wasdue to a fresheningof Thesimulation of restoredunernal hydraulics the tnidwestportion of the basin,and fresh marsh wasdone only to theTerrebonne basin. The rnodifi- preservation Fig. 6b!, The scenariohad negligible cationsimplerrmnted in thisscenario in Terrebonne effectsupon the distributionand amount of swatnp include:! Restorationof natutallevees at 15 sites and salt marsh habitats Table 2!, thathad been destroyed and ! lmprovetrtentofthe Landscape4lodeling inCnasta> Louisiana212 2'l3 E. Re!teeet al.

a! hiodellmplenteuration

h! D>ffcven

W bra

bra

Fig,6. a! Manipulauonsimposed within tnodel to performTerrebonne Restore Hydraulic Conditions scenario. b!. Differenc habitatmap for Terrebonne Restore Hydraulic Conditions scenario vs. Normal Conditions,

StoppingSaltwater latrstslon Scenario habitatchange in coastalLouisiana are j sealevel riseand subsidence, ! changesin theintroduction In this scenario,a barrierto saltwaterinfusion of freshwaterand sediments from the Mississippi into the upperbasin was simulatedfor Barataria, and AtchafalayaRivers, and ! modif!cationsto This levee crossesthe width of the Barataria basin internalhydrology Baumann et al, 1986;Boumans withonly one break in thecenter portion that allows andDay 1993; Cole tnan 1988; Day et al, 1997;Day flowfrom Bayou Perot and Bayou Rigoletts to Little andTemplet 1989; Gagliano et al. 1981;Reed 1995; Lake.The results showed anincrease inopen water Salinaset al. 1986;Turner 1997!. Our landscape Table2! comparedto the NC scenario,About 110 modelsare driven by thesesame dominant regional km' of fresh marshwas lost, 10 km' of brackish processesand were, therefore, sensitive to factors marshwere lost for a totalopen water gain of 88 that affect how land and water surfaces evolve km' Fig,7!, There was preservation ofan additional interactivelythrough time. This meansthat the 35 km-'of saltmarsh. The greatest amount of land models are less sensitive to local factors such as losstook place in fresh marsheslocated in the canal dredging, or natural ones like nutria eat outs northwest.This was duc to increasedflooding or fires. durationnorth of the simulated levee Fig. 7!. When only the western portion of the DlSCUS SION Terrebonnebasin was modeled in the original CELSS study Sklar et al. 1991!,a Ft = 0.&82for Othershave found that the principal regional the calibration run from 1956-1978 and Ft = 0.79 factorsdriving long-term trends in land-lossand for the validation run of 1978-1983 were obtained. Lan~ Modelingin Coastal Louisiana 214

2018 Normal onditions 2018 SaBwater Intrusion

Areal Fmenx tun-! pauerri HahtxaxTape Normal Salvaaxer fresh marsh 393 303 sv amp l7 IOI5 hriickish marsh 336 i 36 [QiiiI salt marsh 1 l9 347 open wmer 4059 4iM uplands 178 l78 lunposedl Fig.7. Resultinghabilat distribution of BaratariaSaltwater Intrusion scenario compaxed vs. Norma] Conditions scenario.

Theresults for bothpresent models which included predictablyinfluenced land elevation and habitat the area modeled by Costanza and his colleagues type. Thesemechanisms had their major impacts axesimilar Table2!. Theseresult s leadus to believe on the marsh communities, thus setting a trend of thatthe preseint tnodel satisfactorily represents the land loss in these areas. This was due in part to Baratariaand Terrebonne systems at a largescale differences in initial elevation, salinity, and water andpmvide with an adequatespatial and temporal levelsbetween upper and lowerbasins. resolutionfor the modelmanagement predictions, CONCLUSIONS The assumptions of the landscape models. paxticularlywith respectto thespatial resolution of Two region al model s that co xnbi ne the waterrouting IOO km'!, limited coxnparisons hydrodynaxnicand biological processesat different betweenadjacent areas, and made assessment of the time and space scales are presented. These iong-termpotential of small projects less than l mechanisticmodels includedfeedback amongfour km'! virtuallyimpossible. The landscapemodels different modules water. soil, plant and habitat shouldbe usedas reconnaissancetools to provide switching!.A -year calibration showedthat thc aninitial estimate of thetype and level of hydrologic processesmodeled are sufficient to explain up to changenecessary to achievea desiredresponse, 85% of the habitat distribution and 95% of the lan Jt' Plantand soil processes atless than l km'scale and water changes. hydrologicprocesses at lessthan IOOkxn' that have ttotbeen xnodeled may produce loca!ized changes. Scenarioresults dexnon strated the importance of increasing water flow into both basins. As these The biological module and habitat switching areas are subject to limited freshvvater inflov s, the algorithxnsfocused on those factors that directly or rate of land loss only increases. 2ts E. Reyas et aI,

The adv antag e of usin g these land scape BOuMAvS, R. M. J. Avo F. H. SKLAR. 1990. A models is that they allow one to evaluate natural polygon-based spat i al P B S! model for processeswith a regional perspectiveand to simulating landscapechange. Landscape investigatecause and effect of the management Ecology,4/3!; 83-97. optionsat any location. Burtntcx,D. M., !, A. ME'voELssoHis,AND K. L. McKEE. 1989. I.ivc standing crop and ACKNOWLEDGMENTS Metabolismof theMarsh grass Sparti na patens as relatedto Edaphic Factorsin a Brackish Financialsupport for this studywas provided marshcommunity in Louisiana. Estuaries by the Barataria-TerrebonneNational Estuary 12!: 195-204. Program BTNEP! throughthe LA, Departmentof CttAnRECX,R. H. 1972.Vegetation, water and SOil EnvironmentalQuality. The authorswould like to characteristicsof the Louisiana coastal region. acknowledgethe assistance of theBTNEP Scientific BatonRouge, Louisiana: Sca Grant, andTechnical Cornrnittee members in reviewing CHAartEctc,R. H. Avv R. E. Coworu:v, ] 979. Common earlyresults. We alsowould like to thankPhillip vascularplants of the Louisiana marsh.Baton Atkinson,Hassan Mashriqui and James Hyfield for Rouge, Louisiana: Sea Grant. thelong hours they spent submitting jobs and post- CHE'vc,R. T., V. CASULLLAvv J. W. GAitrvEn,1993. processing.We acknowledgeMelissa Woods and Tidal, Residual,intertidal Mudflat TRIM! EmilyHyfield for editing and proof-reading. Weare Modeland its Applications to SanFrancisco gratefulto theeditors and two anonymousreviewers Bay,California. Estuarine, Coastal and Shelf fortheir comments and suggestions. Science 36:235-280. COtEMAri, J. M. I988. Dynamicchanges and E.ITERATURE CITED processesin the Mississippi River delta. GeologicalSociety of AmericaBulletin 100. ALAwADV,M. AivoK, AL-TAHA. 1 996.Elevation Data 999-1015. Gathering B arataria-Terrebonne National CovNEa,W. H. AvoJ. W. DAv.1976. Productivity EstuaryProgram BTNEP!. No, Department andcotnposition ofa baldcypress-watertupclo of Civil andEnvironmental Engineering and site and a bottomland hardwood site in a RemoteSensing and ImageProcessing Louistana swamp. American Journal of Laboratory Louisiana State University. Botany 63 I 0!; 1354-1364. BAuMAMis,R,H. 1980. Mechanisms formaintaining Coiner"Eit,W, HJ. W. J. DAv, J. G. GossaiNx, C, S. marshelevation in a subsidingenvironment. Hor Xt tiSON, Asv W. C. Srow E. I 987. Master's,Louisiana State University Vegetation:composition and production. In W, BAVMAvri,R. H., J. W. DAV, ANV C,A. MrLLER, I986, H. Connerand J, W. J. Day Eds!, The Mississippi Deltaic Wetland Survival: Ecologyof BaratariaBasin, Louisiana. pp. SedimentationVersus Coastal Submergence, 31-47!. BatonRouge, Louisiana. Science224: 1093-1095. CosTAisza, R. 1989. 1VIodelGoodness of Fit: a BOEScit,D. F., M, N. JOssELvN,A. J, MEHTA,J. T. multiple resolutionprocedure. Ecological MOaiuS,W. K. Nt~, C. A. SivtnivsTAn,*vn Jrlodelling47: 199-215. D. J. P. SwtFT.1994, Scientific asscssrnent of' CosTAs:zA,R., F. H. SKLAn,Avn M. L. wi-irrE. 1990, coastal wetland loss, restoration and ModeI i ng Coastal Landscape Dynamics. managementin Louisiana Special Issue No. BioScience40!:91-107. 20cd.!. Lawrence, Kansas: Coastal Education DAr, T. ANo R. G. WrEcrar. 1996. Estimation of the and ResearchFoundation. primaryproductivity of Spartinaalterniflora BouiviAvS,R. M. J. AisoJ, W, DAv.1993. Effeeta Of usinga canopymodel. Ecography 19:410-423. twoLouisiana marsh management plans on Day, J. WJ. F. Martin, L. C. Cardoch, and P. H. water and material flux and short-term Ternplet.1997. System functioning as a basis sedimentation.Wetlands 14!:247-26L for sustainable management of deltaic ecosystems,Coastal Afanagement 25:115-153. LandscapeModeling in CoastalLouisiana 216

DAV,J. W, AvD P. H, Trmpunr, 1989, COnSequences MuRRAY,S. P. AwoJ, DoNLEY.1994. Mississippi of Sea Level Rise: Implications from the River Plume Hydrography:Annual Report Mississippi Delta. CoasralManagement 17: annual report No. MMS 94-0028!. Louisiana 2411-257. StateUniversity, Coastal Studies Institute. DuNEAR,J. B., L. D. BRrrSCH,Also E, B. I, KENT, NvstAN, J. A., R. D. DELAum, H, H. RoaEars,AND 1992.Land LossRates; Report 3, Louisiana w. H. PATRtcit,JR. 1993. Relationship between CoastalPlain Report 3 of a seriesNo, GL- vegetationand soil formation in a rapidly 90.2!,United StatesArmy Corpsof Engineers. submergingcoastal marsh.hfarine Ecology GxoUANo, S. M., K. J. MEYER-ARENDT, AvD K, M. ProgressSeries 96:269-279. WtCtcER.1981. Land lOSSin thCMiasiSSippi PENi AND, S., R. BoYD, A'sD J. R. Suraa. 1988. River deltaic plain. Transactionsof the Gulf Transgressivedepositional systems of the CoastAssociation of the Geological Society Mississippi delta plain:a model for barrier 31:295-300. shorelin~and shelf sanddevelopment, Journal He ittttsoN, C, SJ. W. DAY, Avo B, T, GAEL. 1978, of SedimentaryPetrology 58:932-949. RespirationStudies in a LouisianaSalt Marsh, PEnLAjvo,S. ANDK. E. RAMsEY.1990. Relative Sea- Anales del Centro Ciencias del Mar y Level rise in Louisiana and the Gulf of LimnologiaUNAM 5!:225-238. Mexico: 1908-1988, Journal of Coastal Hemmed, C, S R, L, Wi rrsri Axn J. W. DAY. 1988. Research6!:323-342, Simulation Models of Coastal Wetland and PERRET,W. S., B. B. BARRY, W. R. LATApu:,J. F. Estuarine Systems: Realization of Goals. In PoLLARD,W. R. Mocx, G. 8. AoKn s, W. J. M. S. W J. Mitsch, S.E. Jtsrgensen Ed.!, GAtnitv, ANDC. J. Wttrrn. 1971, COOperative Wetland Modelling, Developments in Gulf of Mexicoestuarine inventory and study. EnvironmentalModelling 12 pp. 67-96!. ln Louisiana, Phase 1, Area Description pp. Elsevier Science Publishers B,V. 31-175!, Wildlife andFisheries Commission. HowEs,B. L., J W. H. DAcEY,AND J. M. TEAL. 1985. Htu,S, R., R. D. DELAuNE, AND W. H. PAm!Ctc, Annual carbon tnineralization and 1987, Response of Spartina patens to belowground production of Spartina increasing levels of salinity in rapidly alterniflorain a New England salt marsh, subsiding marshesof the Mississippi River Ecology66!:S95-605, deltaic plain. Estuarine,Coastal and Shelf HotvEs,B. L., W, H. DAcav, ANDD, D. GDEHRtvcEa, Sci ence 24:389-399. 1986. Factors controlling the growth forin of Rr~, D.J. E. 1995. Status andTrends of Hydrologic Spartinaalterniflora: feedbacksbetween Modification, Reduction in Sedi ment above-groundproduction, seditnent oxidation, Availability, andHabitat Loss/Modification in nitrogenand salinity. Journal nf Ecology74: the Barataria-TerrebonneEstuarine System 881-898. No. BTNEP No. 20!. Barataria- Terrebonne KJRBY,C. J. ANnJ. G. Gossrttv'.. 1976. Primary National Estuary Program. Pmductionin a LouisianaGulf Coast Spartina REYES, E., J. W. DAY, AsD F H. SKEAR. 1994. alternifloraMarsh, Ecology 57:1052-1059. Ecosystem Models of Aquatic Primary LAERESnartt,W. K., D. L. URaAts,D. P. COFFn.,W, Production and Fish Migration in Laguna de J, PARTDN,H. H. SHtjoART,T. B. Ktac>nvER,Avn Terminos, Mexico. ln W. J Mitsch Ed.!, T. M. SMtrH. 1993. Modeling vegetation Global Wetlands.-Old IVor d and Nev pp. suucture-ecosystemprocess interaction across 519-536!. Amsterdain: Elsevier Scientific sites and ecosystetns,Ecological Modeling Publishers. 7!:49-80, REvEs, E., M. WHiTE, J. MARTtv',G. P. Krvtp, J. DAY, Liu, J., F. W. CueaAOF.AND R. H. PuLL.tAM1994. AvnV, ARAvAMurisRAv,In Review. Landscape Ecologicaland econotniceffects of forest Modelingof CoastalHabitat Change in the landscapestructure and rotation length: Mississippi Delta. Ecology. simulation studies using ECOLECON. RoeERTs,H. H. 1997. Dynamic changesof the Ecological Economics 10!: 249-263. Holoccnc Mississippi River delta plain: the 217 E. Reyea lt al.

deltacycle. Journal of CoastalResearch 13: WHrTE,M. L, 1991, Spatial Modelling in Coastal 605-627. Louisiana.In S. Mathies Ed.!, DataInventory R, D. DELAt!rate, x!ro W. H. PaTRrcK Workshop,5 pp, 367-376!. Thibodaux,LA: JR.1986. Changes Occumng Along A Rapidly LouisianaState University and BTNEP, SubmergingCoastal Area: Louisiana, USA, WrnTE, M, L., J. F. MprtrlN, E. Rrtvr:s, G. P. KExrr, 3, JourrrraI of CoastalResearch 2!:269-284. W. DAY,V. ARAVAMtrrrlAs, 3. N. ScHAYDA,Ar'D SrNGN,V, P ~Nt! 'V.ArtAVAMt!rrrAN 1995. Errors of H. S.MasHRlqVI. 1997. LandsCape SimulatiOn Kinematic-Wave and Diffustion Wave Model Upgrading; spatial-ecological Approximationsfor Time-Independent Flows, modeling of the wetlands of Barataria- Water ResourcesManager!rent 9!:755-782. Terrebonne BTNEP Final Report No. 31!. SOLAR, F. H. w~n R, CosraNza. 1991, Thc CoastalEcology Institute, LSU. Developmentof DynamicSpatial Models for WrnTE,M, L., T. M~wEt.L, R.Cosrs NzA,~o T. W, LandscapeEcology; A Reviewand Prognosis, DovLr.". 1991. Fcosystem Modeling of ln M. G. Turner and R. H. Gardner EtLs.h BaratariaBasin, Louisiana: Utilizing Desktop Quantitative Methodsin LandscapeEcology Parallel Technology No. American Water pp. 239-288!. NewYork, New York; Springer- Resources Association. Verlag, WrsExs*rs,W. 3., E. M. Swr:wsow,ANo J. PowErt,1990. SrtLxrt,F- H., R. CosTxvzw,aNn J. W, Dxv, 3!t. 1985. Salinity Trends in Louisiana Estuaries Dynamic SpatialSimulation Modelling of Estuaries 13!;265-271, Coastal Wetland Habitat Succession. Wu, J, G, !twoS. A. Lavrv. 1994. A spatial patch Ecological Modelling 29:261-281. dynamic modelingapproach to pattern and Sr'!ut, F. H., 1Vl. L. WHrrE, xNo R, CosTmz!s. 1991. process in an annual grassland. Ecological The Coastal EcologicalLandscape Spatial Monographs 64!;447-464. Simulation CELSS! Model: Users Guide and Resultsfor theAtchafalayafI'errebonne Study Area Open File Report91-04 No, National We t 1ands Re search Ce nt er, US F ish and Wildlife Service, Trroxtsos, D. J, 1995. The Seasons, Global Temperature. and Precession.Science, 268 April !, 59-68, T+F>, R. W. 1993. Field Guide to Coastal Wetland Plants of the Southeastern United States. Arnherst.Ma~~achusetts: The University of Massachusetts Press. TuR~Ert,R. E. 1997. Wetland loss in the northern Gulfof mexico:multiple working hypotheses. Esrrusries 20 I !:1-13. Vrssrst,J. M., C. E. SxssErt,R. H. CrrABREcK,Avn R. G LtNscoxsar:.1996. Marsh Vegetation- Types of Barataria andTerrebonne Estuaries, No, 1.$ O'- C EI -96-11 !, B arataria- Terr ebonne Nat.ional EstuaryProgram. WELts,3. T. 1996. Subsidence,sea-level rise, and wetlandloss in thelower Mississippi River delta,In J. D. Millimanand B, U. Haq Eds.!, Sea-LevelRise and Coasral Subsidence pp. 281-3I I ! Netherlands:kluwer Academic Publishers. Salt Marsh Plant Colonization, Growth, anal Dominance on Large Mudflats Created Using Dredged Sediments

C. EDwARD PROFFITr' AND JENNJFERYOUNG Louisiana Envirr>rtmentalResearch Center, McNeeseStare University, P.O. Box 90220, Lake Charles, LA 70609

Present addresses: 'Author i'orcorrespondence; U.S. GeologicalSurvey, Biological Resources Division. NationalWetland Research Center, 700 CajundomeBlvd., Lafayette,LA 70506 TEL: 3 l 8-266-8509; FAX: 318-266-8592; email: edward proNttC~usgs,gov

'LouisianaDeparttnent of NaturalResources, Nicholls State University, l07 Mead Hall, Thibodaux, LA 70310

ABSTRACT: Large mudhats up to e200 ha! created from sedimentsdtmlged from the Calcasieu Ship Channel are being colonizedby marsh halophytes.We comparedcolonization and gro»1h by dominant marsh vegetationat edgeand interior locationsof createdsites ranging in age from 6 mo to 14 yr and natural referencemsrshes. Unvegetated arcs declinedrapidly with ageof cited marsbes interior locations: 99.7% io the6 mo old marsh, 428% in tbe3 yr old marsh, and 0.1% in the 14 yr old insrsh!.This change occuicdprhnariiy as s resultof rapid colonizationand vegetativegrowth by smooth cordgrass, Spaesirsarzlterriijlortr, in sll sites. The first year's colonizationwas mainly by raMng of mats of roots and rhizoines,but vertkal growth of buried plants was also important m sotnemarsh edge locations. Seedlingscontributed to marsh grassrecraaitment in tbe secondand third years in the two youngestsites. Sptzrtinaalternijlnr» dominatedor ccwiomiuatedin our 1996 survey: a! 14.4% edge! - 2.7% interior! of site III created in 1996 and surveyedabout 6 inontbs after de-watering,h! 82.7% cdge!- 279% interior! of site 11created in 92993,and c! l00% edge!- 46 8% {interior! of site I createdin 1982, d! 100% edge!- 922% {interior! of referenceinsrsh 1, and e! 99% edge! and 100% interior! of referencemarsh 2. Other marshgrass species dominated or co- dominstedfess area. Dombtanceby S. pasensranged from 0- 15'% {createdmarshes! «nd t - 5.8% {rcfescncemarshes! depending on location andsite. Donunanceby Disrichlisspicutrr ranged from !- 600% crestcd inarshes! and 0 - 2.0% referencemarshes! depending on location and site. The annualsucctdent Ssiricorniu bigekvii exhibited thick growthbetween patches of grassin many portions of the interior of site IL Spartirraalterrtiflons wss shorter aad hsdgreater raxnet densities in siteI, the oldest created marsh, relative to the other tnarshes. Plant height and biomass was greatest in site Il. There wasuo significantdifference in plant heightand biomassdue to locationwithin marshes edge vs interior!. The numbers of floweringstems of S. altensiglorowerc not diferent in site II and tbe natural refeicncesmarsh 1. However,there wasno flowerin in siteI the 14 yr old createdmarsh!. The causeof these difl'erencesamong populations could be either environmentalor genetic. Spartina alterniflora cover reduced below-canopy air temperaturesdming summer,but had no significant effect iu other setisons.Sediment organic matter generallyincreased with the age of the created marsh.

Key wordsand phrases: salt marshcreation and restoration, marshp!ants. dredge sediments.Spurtirtu aherrtiflora, Spartina paterts, Disrichlis spicatu, Sulicorm'ubigelovii

Fromthe Symposium Recerrr Reseurck irr Coastall~utriarra. ktaruraiSysrern Function und Response ru Hurrrurr /rrfluertee. Rozas,L.P., 3.A. Nyman, C.E, Proffitt, N.N. Rabalais,DJ. Read,and R.E. Turner editors!. 999. Publishedby Louisiana SeaGrant Col!ege Program.

218 219 C.E. Proffttt and J. Young

INTRODUCTION noted above. Studies of' colonization, growth, and speciesinteractions are required to assessthe rate Restoration of marsh structure and function is of marshdevelopment and patternsof succession. a criticalcomponent of the efforts to offset wetland We studiedthe colonizationand vegetativegrowth lossesin coastalLouisiana. One restorationmethod of thedominant grass species Sparri na alterrtiflrzra. involvesthe creationof rnarshhabitatin openwater We also characterizedthe percent cover of afl areasby usingseditnents dredged for either that abundantgrass species S. alternifloru,S. paterts, purposeor asa by-productof navigationchannel and Dr'sriehlisspicata L,! Greene! in 3 created maintenance. Marsh creation has sometirrtes marshesof differentages and 2 referencemarshes includedplanting one or more specieson the in order to assesspatterns of succession and dredgedsediments, However, in manyinstances, dominance. This effort was part of a larger large bare mud flats are created under the collaborative suite of studies that also included assumptionsthat plants wil I colonize"naturally," analysesof the geneticdiversity of S. alrerntJfara that the ensuing tnix of dotninantspecies is andthe levels of metaland organ ic contaminants in determinedtnainly by elevation and hydrology, and sediment and biota, thatthese "tnarshes"will have atl or mostecological functionstypically associatedwith tnarshes of that METHODS groupof species.These assumptions have not been fully tested,especially in largearea sites. Site Descriptions

Environmental gradients, such as tidal Three large flats >40-200 ha! of dredged inundation, salinity, and soil anoxia, have been sedimentswere created by the U S. Army Corps of shown to affect tidal marsh plant growth and Engineersin 1982, 1993,and 1996 at the Sabine survival Ponnamperurna 1972, Mendelssohn et al, NationalWildlife Refugein southwesternLouisiana 1981,King et al. 1982, Nixon 1982,Mitsch and duringmaintenance dredging of the CalcasieuShip Gosselink1986, andmany others!.Interspeciflc Channel, All sites used in this study were on the competitionbetween dominant salt marsh grasses eastside of statehighway 27, southof thecity limits alongthe gradients establis,hes the typicalzonation of Hackberry, LA «nd north of the refuge patternsnoted in high and low intertidalmarsh headquarters.The flats werecreated by pumping a settings Bertness and Ellison 1987, Bertness 1991!, slurry of dredgedsediments and water into areas In low marsh, flooded soils are often anoxic and that hadhistorically been marsh, but which were the ability of Spartina alterniflara Loisel to openwater with occasionalmarsh islands at thetime oxygenateits rootsand rhizophereis a key factor of restoration. All of the created marsh sites have allowingthis speciesto becomedominant Gleason contaimnent berrns on one or two sides and are 1980,Mendelssohn et al. 1981!, High marshsoils contiguouswith naturalsalt marshor openwater aresometimes more oxygenated and consequently onother sides. Marsheswere open to floodingfrotn are often lessstressful in tertns of this physical either the Hog Island Gully canal site II and factor.Sparrina alternigora can colonize and grow referencemarshes I and2!, old Hog Island Gully underhigh marshconditions. However, it will be and West Cove site III!, or the Calcasieu Ship excludedwhen Spartinaparens Ait,! Muhl., a Channeland Long Bayou site I!. Our studywas strong competitor for space, is also in the site initiatedprior to constructionof site Ill in 1996. Bertness 1991!. Thus, this site was sampledfor certain variables percentcover, clone size, and sedimentorganic The southwesternportion of thc Calcasieu content!but was not includedin the quantitative estuarynear West Cove is strongly influencedby analysesof growth,biomass, etc, Naturalmarshes oceanicwater entering through the Calcasieu Ship adjacentto the flats were used as referencesites. Channel.Consequently, marshes on the east side Reference marsh 1 was located on the north side of of highway27, whereour studytook place, are theHog IslandGully canalto the west of the site II dominatedby halophytessuch as the marsh grasses createdtnarsh. This referencemarsh was sampled Salt MarshOeve!oprnent on Dredt!edSediments for cover, stern density, height, and flowering werecorded numbers of stems,!ength o fevery stern whenevercreated rnarshes I and II weresampled. asa measureof height, nutnbers offlowering stems, We establisheda secondreference marsh following andcollected all aboveground material for biomass creation of the l996 site. Thi» reference marsh 2 determination dry mass after drying at 70" C!. A! so, waslocated on the south side of the HogIsland Gully we recordedwater depth and salinity at eachplot- canalimmediately to the westof site III. Air temperaturewas measuredabove and below the S. alterniflor canopy Qunnhtativeaaa!yses of biomassand plant size and environmenta! conditions Biomass,height floweringand stem density datawere analyzedby repeated-measuresor two- In April 1996 we established stationsfor study wayANOVA asappropriate using Systat, Repeated- attwo locations "edge" near open water or adjacent measuresanalysis was used for temporaldata when marshand "interior" further from open water! in theassutnptions of thetest were met. Differenc s in two created marshes site I created in ! 982 and site air temperatureabove and below canopy was tested II createdin 1993! and a reference marsh reference usinga pairedt-test. 1,adjacent to site!I!.In site I andreference I, edge locationswere within 2 mof openwater and interior Percent Cover Determinations plotswere 50 m inland.Using 50 m fromopen water for interiorplots sufficed because these tnarshes We ran a minimumof 3 replicatetransects t 4 werean! y approxitnately150 - 200 tn wideand thus rn wide! across each marsh in all created and interiorp!ots actually represented marsh interior, reference marshes. We measured the distances of However, this was not the case in the much-larger open ground or water! and all vegetatedpatches siteII. This site wascontinuous with existingnatural fa!!ingon the transectline, The dominantspecies, marshon its westernmargin. As "edge"locations or co-dotninantspecies mixture assessedby visua! in this large 00 ha! site, we picked plots of estimationas approxitnately equal spatial coverage !, vegetationthat were c!ear!ygrowing on dredged was recorded for each section of vegetation. The sedimentsand were closestto the borderingnatural percentcover was calculated from the meansof the marsh.For the "interior" plots, we chosea location replicatetransects in eachmarsh. nearer marsh center about 400 m from the edge whichappeared representative of thetypical interior Sediment Organic Matter of the site. Thus, for all study marshes,interior locations were situated between about !/4 and !/2 As one metric of long-tertn effects of of the distancefrom open waterto the mostdistant vegetationon soil structure,we conducteda one- pointacross the marsh. time samplingof the studyplots for sediment organiccontent. Litter was removed from surface Three permanent p!ots were establishedand upper5 cm!soil samples and the stx!imcntwas markedwith PVC polesat eachlocation. In siteII, driedto constant temperature at7P C. Drysampies whereS. alterniflnra often existed as spatially wereusual ly veryhard and were crushed and ground separatedcircular patches, plots were locatedin priorto determiningorgaruc content via loss on separatepatches of the grass. The patch radius at ignition00' C for 4 bours!. studyinitiation in April 1996 rangedfrom 3,06 to 7.49 m. Cominuousswards of grassexisted rather RESULTS thanseparate small patches,in site I and reference marshes.In thesesites, plots at a location edge or Knvironmentn! Conditions interior!were situated 50 m apart. E!evations in thernarshes range from 0 to + !,7 We sampledeach plot quarterlyover ! year. m with 0 defined as elevation at marsh-open water Ten quadrats 5 x 25 cm! were !ocated by interface C. Norman,unpublished data!. In siteI I, haphazardtoss at eachsampling. In eachquadrat elcvationsatthe edge plots ranged from o-! 5 to G,24 221 C.E. Pmffittand J. Young

m andfrom 0.31 to0.46 m atinterior plots Because Historicdata provided by the SabineNational of logistical constraints,elevations were not Wi!d!ife Refuge show considerablewithin-and- gathered«t individual study plots in the other among-yearfluctuations in salinitiesin the Hog marshes.Whi!e thesites were relatively flat, there IslandGu!!y Cana! adjacent to restoredsites l!and werctopographic highs in sitesII andIII whichwere III andboth reference marshes Fig. 1!, From 1991 probablyassociated with theplacement of discharge - 1997,the mean 4 1 S,I:, salinity was10.7 i 0.51 pipeopenings during sediment deposition. mg1 ' n = 2, andthe range was 0.7 to 23.4mg 1 ~.The salinities of groundwater and surface waters Water depths varied seasonal I y, am ong in ourstudy ranged from A! to 32.0i 8.0 mg 1' marshes, and between !ocations within marsh es. The TableI !,and generally tracked patterns of change driesttime was the first co!lection spring !996! with of salinitiesin the canal. Salinityvalues in site I overamean depths to groundwater below surface the yr old createdmarsh! were typica! ly a little valuesare indicated by negative sign! in created sites higherthan in other sites,This site is nearestthe of -66.3 k 2.62 cm Site II and -I8.8 4 2.1! cm in CalcasieuShip Channel. SiteI. !nreference marsh I therewas standing water at this co!lection ,4 2 2.20 ctn!. In the winter of At higher elevationsin two of the created !997,all marsheshad standing saline wa ter S ite II tnarshes sites II and III! surface sediments, 5,611,91 crn, Site.0% 1.89cm, Referencemarsh especiallyin unvegetatedareas, werc often hard, 1 8.92 1.90 crn, andReference marsh 2 18.8 2 2.10 dry,and cracked when there was no standing water cm!. Waterelevation data were notco!!ectedin the or recent rainfa, Under these conditions, surface youngestrestored marsh site III! because it was sedimentsalinities were substantially higher than notde-watered enough to al!ow walking access for thoseof the groundwater oftcn 20-60 cm below samplinguntil late intothe study, sedimentsurface!. Sedimentsalinities from site!I,

25

20

15 IÃ > io

tn

0

Fig.l. Historicbottom salinities frotn 1991 - l997in the Hog Island Gully Canal are shown. On the x-axis. the year labelis p!aced atthe start of a particularyear.Data collection varied but points were gathered approximatety bitnonth!y.Data were provided by R,Walters of theSabine National Wi!t!hfe Refuge U.S. Fish and Wildlife Service!. Salt MarshDevelopment on DredgedSediments 222

'Ihble1. Quarter!y measuresof mean + I S.E.! dredged sedimentswas not thick, buried S. saliaitiesin mg I' of standing or ground water alrerrtiflaracolonized by vegetativegrowth and 2! areshown for eachstudy site nM!. ND indicates overmuch of thesite numerous small generally data not collected. Zero values indicate fresh 0.5 mi! patchesof S. alrrr7iiflorawith intact roots water within the limits of measurement. and rhizomes rafted in with tides and storms, Sparrirtaparerts colonized by rafting and 0, spicara Mean Salinity Mg 1' I S E ! apparently by seed over those early months followingsite de-watering. However, relative to S. Site II S it c I Reference alrernillora,these latter two species did not co!onize yr old! 4 yr old! Marsh I extensivelyin termsof n.umbersof patchesand total area covered, Apr 96 22.0 .0! 32.0 8.0!

Jul 96 21.7 .5! 20.0 .0! 16,3,4! Percent Cover

Vov 96 19,3 .1! 21.8 .8! 17.2 .3! The unvegetatedarea in createdmarshcs Mar 97 2,2 .1! 0 0 decreasedwith tnarshage and increasedwith distancefrom open water Table2!. Datafrom sites II andIII createdin 1993and 1996,respectively! measuredin July 1996 aftermixing equal volumes showthat colonization was more rapid near the edge of distilledwater and dry sedimentand allowing 5 of thc marshcompared to theinterior of the marsh. hrto cometo equilibrium,were 50.2 2 4.1 Mg 1' Field observations in .site III suggest that a n = 5!. Sediment from reference sites, site I, and substantialproportion of thiscolonization near the near-open-waterareas in siteII that wastreated in edge was from vertical growth of buried S. the samemanner, had much lower salinities that alrerniflora. Both referencemarshes had nearly weresimilar to those.of groundwater, 100% vegetative cover Table 2!. Sparrina alrerniflora was the dominantor co-dominant The % organicmatter differed amongsites speciesin al! sites Table 2!. At edgelocations one-wayANOVA, P =0.0005, R' = 0.876!. Values within 2 m ofopen water! in createdsites, the area from site I created 1982! were not significantly coveredby S, alrernifloraalone or asa co-dominant di ferent from the reference 1 marsh 2. I and 9.5% increasedwith ageof thesite from14.4% site III, respectively!,The two youngestmarshes sites II created1996!, to 82.7% siteII. created1993!, and andIII! had less sedimentorganic rnatter .7- 100% site I, created 1982!. At interior locations 7.3%! than the reference marsh. Reference marsh 2 ! 50 m fromtnarsh edge! in the createdsites, area hadthe highest sediment organic content 9.3%!. with S. alterrtiflora as sole-or-co-dominantalso increasedwith site agef'orn 2.7% siteIII, created Air teinperaturcs above and below S. 1996!,to 27.3% site ll created1993!, and46,8% altrrrtifloracanopy were significantly different only site I, created1982! . Edge locations of the 2 duringthe summer t-test,P = 0.0005!. The overall reference marshes had 99 - 100% cover by S. meantemperature was 3.9" C cooler canopy in alrerniflor Table2!. DominancebyS, alterrt rflnra sulilinen in interior locationswas typically less than at marsh edge,but showedthe same increasing trend with Field Observations of Initial Colonization increasing age of createdmarsh; 2.7% site III, 27,3% site II, and46.8% site I, Referencemarshcs had 92.2 Field observations of site III, which was - 100% dominaceby S. alrerrtifloraat interior created and became de-watered during our locations. investigations,indicate that grasscolonization occurred within a few months of sediment de- In interiorlocations of the oldestcreated marsh wateringand as a result of two mainmechanisms: site I! the areadominated or co-dominatedby the 1! in thelow intertidalzone, where the layerof highmarsh grass D. spicata was about thc same as 2' C,E. Proftlttand J. Young

Table2. Percentcover by the ruajorhabitat types in createdand naturalreference marshes. Values are tucanaof n=3 transectsper locationin eachmarsh and as suchmay nottotal to 100k. Mixed speciesgrou.pings Indicate numerical dominance by thespecies included. ~~ indicates open water at hightide ln tMsmarsh. Note that Salicortust bigelovii Is an annualand in thewinter, this portion of the habitatis aho bare ground with standingdead plants!. MIxed dotninancewas recorded when species coveredaplmosimately equal proportions of a patch.Ages of created sites at timeof sasnplingwas site Ill 8 yr!, siteII 0 yr!, and siteI I4 yr!. In eachmarsh, "edge" denotes the area from water'sedge to SOm Inland; and, '1nterior" denotesFrom that point on to the approximatecenter of the marsh.

Created Marshes Reference Marshes

Site III Site II Site I Ref 1 Ref 2 edge interior edge interior edge interior edge interior cdgc interior

Bare Ground 85.1 97,0 2.4 42.8 0 0 0 0 0 l~' 0 Sahco rrrirs bigelovii 0 0 2.5 27,5 0 0 0 0 0 0 Spartinaalterrr rflanr 14.4 2.4 68.3 27.3 92.0 22.0 100 92.2 71.0 98,4 Spartina par eris 0.1 0.2 4.0 ],7 0 12,0 0 58 0 0 Distichlis spi cata 0.3 0.4 8.6 0.7 0 28.8 0 20 0 0 1vafrit tescerrs 0 0 0 0 0 0.2 0 0 0 0 S. alter7r.+S. patens 0 0 10,5 0 0 0 0 0 9.2 0 S. altern.+D, spicata 0 03 3.9 0 8.0 9.5 0 0 18.8 2,6 S alterrr.+B. frutescens 0 0 0 0 0 1,5 0 0 0 0 S. patens+ I vafrrr tescerrs 0 0 0 0 0 250 0 0 0 D. spicata+/. frtrrescerrs 0 0 0 0 0 7,6 0 0 0 0 D. spicata+B,frutescens 0 0 0 0 0 O.e 0 0 0 0 S paterrs+I.fry.+B.fru. 0 0 0 0 0 1.3 0 0 0 0 S. alt.+ D. spic.+[.fru. 0 0 0 0 0 138 0 0 0 0 +B. fru. that dominated by S. alterrtiflora Table 2!. This The annual Salicorrria bigelovii dominated speciescovered touch less area in other created ungrassedareas and displayedvery robust growth marshes,0 - 12.5%depending on marshand location duringspring and summerin the interiorof site II within marsh and 0 - I 8.8% in reference marshes Tab!e2!. Salicorrriabigelovii individualsgrowirlg Table 2!. In created site I, where it was tnost within grassedpatches were much smaller and abundant,the area dominatedby the third most occurredin low densities field observations!.It was abundantgrass species, S. parens, cvas about 3x less possibleto findoccasional S, bigelovii in theinterior thanthe greatestdegree of coverby D. spicata.In of the older site I created 1982! mixed in with the referencernarshes S, parens dotninated 0 - 9,2% of grassesbut plantsdid not attain the high density site area Table 2!. Patchesdominated by tnixturec and large stature seen in site II. In site III created of speciesincreased in frequencyof occurrenceand 1996! there was very little colonization by S. totalsite areacovered with increasing age of created bigelo vii the first year 996-1997 data! and no part marsh,but were iessprevalent in both reference of the site hadany degreeof "dominance"in terms marshes Table 2!, of coverby this species Table 2!. However.our Salt Marsh Development on Dredged Sediments 224 field observationsthc following year summer of 1998!indicate substantialcover by S, bigelvviiin partsof s ibeIII resultingfroin extensive colonization byseeds/seedlings in obvious natural low elevation 160i drift zones where seedscame in on the tide and in 106i apparent seedling shadows around the few 66q individualsthat did colonizethe firstyear. I-urthcr WNI96 lar99 field observationsin the springof 1999 indicatea veryhigh density of S, higelnr ii nowin siteIII. Thus, Sr99II thepattern of extensivecover of non-grassed portionsof interiorareas by S. bigelvvr'i recorded in site II created1993! is being repeatedin site III created 1996!. 'Ll DetMItlesof $. allerrtiflo ra 19996

nsf9 Spartirraalrerniflvra stem densities were 1.5 900 - 2 x greaterin the oldestcreated marsh site I!, thanin site Il, or in the referencemarsh 1 Fig. 2!. t7 9IXI Repeated-ineasures ANOVA showed that I 66 Iaa differencesamong rnarshes was significant P = 0,0005!. There were no significantmain effect ~ 9I96 atm96 I9996 ~ pr97 differencesbetween edge and interior locations Figure2. Rametdensines stems/[25 x 25 cm!! of 5. withinmarshes PW.693!, althoughthe significant alterniflvra are presentedfor restoredmarshes Sites l interaction P = 0.022! indicatedthat densities varied and II and the natural reference marsh by marshedge differentlyamong locations in thedifferent rnarshes and marsh kntenor locations. The x-axis is samphng Fig. 2!. times Apr. 96 Mar. 97!. Valuesare mean~+ l standard errxrr. Stem densities also varied significantly seasonally P = 0.0005!. Bothcreated sites I andII indicatedby a timex siteinteraction P = 0.0005t. showed continual increases in density over the The incanheights in siteI werealways less than 40 courseof one year, while densitiesin reference cm and did notvary substantiallyover the year Fig. marsh 1 irtcreased,and then decreased,a both edge 3!. Heightsin referenceinarsh I rangedfrom about andinterior locations. This resultedin a significant 60 - 80 cm and both edge and interiorlocations time x site interaction P = 0.0005!. showedgeneral decreases in heightover the year, Heightsin siteII rangedfroin about55 - 100cm Heigbtsof S. akernifbro and fluctuated dramatically over the year Fig. 3!. The tallestplants were at theinterior location of Stem length height! was significantly site II and were situatedfar from open water.The erent P = 0,0005!among marshes Fig. 3!. Site mean values exceeded ] 00 crn and some individual I, the oldest created site created 1982!, was stemsa!orig the outer portion of circularclones of dominatedby short-formS. alrerniflora,while site the grasswere 2 m long u.npublisheddata!. II created 1993! and reference marsh 1 were populatedby muchtaller S. aherniflora. Therewerc AbOve GrOttnd BiamaSSOf S. aLterr6iflorrr no significantdifferences P = 0.330! in height at edgeand interior locations within rnarshes Fig. 3!. Aboveground biomass is generallydependent Heightvaried significantly over seasonduring the on bothstem density and height. In ourstudy sites, studyyear P = 0.0005! and differencesamong differencesin plantheight tended to dcterininethe marshesin the pattern of change over time is differencesin biomass.The aboveground biomass

Salt MarshDevelopment on DredgedSedrments 226

dominant or co-dominant over 27.3 % interior oj location!and 82.7% edgelocation! of' the marsh Table 2!.This speciesdominated or co-dominated 46,8% of marsh area a interior locations, and occurredin near mono-specificstands ai edge locationsin site I surveyed 14 years after site creation!. In the two reference marshes, S, alrerniflorawas dominantor co-dotninant over a tninimum of 92.2% of site area regardlessof nearnessto open water Table2!,

Fig.5. Mean I S.E.! numberof floweringstems are Despite dominance by S. alrerniJl ora, the dry shownfor theearly Fall 1996collection, Values are conditionsfor portionsof the year in the interior depictedby differentmarsh site I, site II, andref I! portionsof sitesII andIII appearedconducive for andlocation within marsh. Solid bars are edge the establishment of high marsh species or locationsand open bars are marshintenor locations. Them was no flowering in site I. transitional species,In site III surveyedfor percent coverwhen 0.5 yearsof age!,occasional Salicornia bigeloviiexisted; and, in site II at 3 yearsof age! growth resultedin a high percentcover by this therewas a densespring andsummer 996! cover speciesin the marshescreated in 1982 site I! and by this succulenthalophyte over much of the non- 1993 site II!, For the interiorportions of the created grassedareas in theinterior of thedeveloping marsh sites,this tnechanism probably accounted for tnost Table 2!. This species is an annual and its of the first year colonizationby S, alrerrgiflora. colonizationwas by seed.In the youngestsite site Depositedsediment covered portions of theexisting III, created1996!, S. bigeloviiwas absent in the tnarshalong site edgesthat were continuouswith summer of I996 which is consistent with the fact existingnatural marsh and numerous patches of S. that the site was not yet dc-wateredwhen seeds altemifloraoccurred via vertical growth fmm buried wouldhave been establishing. In thespring-surruner rootsand rhizomes site III, field observations!. of 1997 we observed widely scatteredindividuals of S. bigelovii in site III. In 1998 and 1999 there has Sexualreproduction may havecontributed to beenincreasing dominance bv thisspecies over most colonizationof the site in lateryean. We observed of the non-grassed area of this site field a largeseed set by S. allerniflorain thefall of 1996 observations!.This speciesgrew well in bare areas insite H createdin 1993!and a numberof seedlings in sitesIl andIII, but wassparse and muchsinaller thefollowing spring. It is possiblethat some patches when it occurred within grassed patches. We of thisspecies also became establishedthrough conjecturethat the annualS. bi gelovii leadsa rather seedlingsthe previousyear sincethis site was 3,5 ephemeralexistence in thesecreated marshes, with yearsold when we first inade field observations.In its dominanceincreasingly litnited over the years siteIIL therewere seedling "shadows" around many by spreadof the perennial grasses. clonalpatches in the springof 1998 about2 years aftersite creation field observations!. Typicalhigh tnarshgrasses S. parens and D. spicara! only accounted for at most 14.5% and 8.6% Rapid vegetativegrowth of S. alrerniflora areacover respectively in site II and <0.5% in site resuhedin its dominancein all sitesregardless of III. Wherethese species have colonized, the patches theelevation and distance to marshedge Table2!. appear to be healthy and growing well field Sparrt'naalterriiflora was dominant or co-dotninant observations!. However, there has been little over2.7% interior location! and 14.4% edge colonization,We hypothesizethat this may be a location!in siteIII whenit surveyedat 6 tnonths result of a paucity of rafting propagules,or viable aftersite creation. Spanina alterniflora in site II seed,or both, BecauseS. alrerniIlora generally surveyed3 years after site creation! was the single dominatesnear water's edge in marshesin this 227 C.E. Protfittand J. Young

portionof theSabine National Wildlife Refuge,this significantlylarger height and biomass!, lowerin is thespecies most likely to breakfree and raft into ramet density, and flowcrcd during the study. a newlocation. The other grass species are further Whether this is some function of sediment or other removedfront the waterand often S, alter jlora physical factors or is a difference in doininant occursbetween them and openwater, This result genotype s!is not yet known.We did observebut wouldtend to reducethe frequencyof rarnetsof did notquantify, more frequently water-loggal soils thesespecies breaking free, washing out of a marsh and apparentreddish iron depositsin site I where andfloating to newsites. Although we observedall short-formS. ai/erniflora dominated.Long tenn 3 grassspecies flowering, we haveno data on the studies of these sediments, during years of abundance of viable seed. We have observed S, compactionand/or other structuraland chemical afrerniflaraseedlings in both sites11 and III, but changes,and their effectson vegetationmay be not seedlingsof D, spirara andS, patens. If seed needed to shed light on this question. Also. set or seedlingsurvival is low,this too could limit transplantexperiments among sites and locations colonizationof thesehigh marshgrasses. These couldaddress questions of growthform. observationssuggest that further studies of seedset, viability, floatation, and colonization of the Full assessment of marsh function involves petenniaIgrasses «rc needed to resolvequestions ineasurements of such diverse items as sediment regardingcolonization by thesespecies. biogeochemistryand nutrients, productivity, food webs,population biology and genetic viability, and While the physicalconditions of the interior successionpatterns Such an undertakingwas nat of sitesII and Ill and studiesof competitive withinthe scope of thisstudy. However, in addition outcomes Bertness 1991! suggest that S, parens and/ to biologicaldata on colonizationand growth of the or D. spicarashould eventuawy come to be the doininantplant species,we did evaluatea few dominatepertenial grasses in interiorportions of physical parametersthat are affected by plant sitesII andI II. therapidity ofcolonization. degree communitydevelopment, These variables were: air af dominance,and robustness of growthof S. temperature above-and-below-canopy as an alterrufloraand the slowness of colonization by S. indicator of how plant cover can affect rnicro- parens and D, spica a in these locationsindicates environmentalconditions; and, sedimentorganic thatthis process may take many years. Interestingly, matteras a metric of long-term changespaJtially dominancebymixed-spec iesassemblages increased causedby plantgrowth and decay with «geof createdmarshes, and was lower in the presumablymuch older reference marshes Table The sedimentorgailic rnatter content of the 2!, Thepreponderance of patches dominated by oldestrestored marsh site I! was not significantly mixturesof speciesin siteI suggeststhat created different from tha in the reference marsh I, marshesof intermediateage thissite was 14 years suggestinglong-term effects of ma+.h grassgro wth old whensurveyed! may be fertilelocations for anddecomposition on soil structure.Sediment from futurestudies of species-speciesinteractions, theyounger restored sites si tesIl and III! hadabout 2/3 as much organic material%! as in the reference Contpstrisonof Marsh Structureaud Function inarsh1 9.5%!. Valuesfor organic matter in surface sedimentsof the reference marsh 2 9.3%! werc Thetwo created marshes sites I andII! studied similar to results froin a marsh terracingstudy for suchvegetation factors assize, growth, biomass locatedc0.5 km to the south L.A Departmentof and reproductionwere both dominated by S. Natural Resources Coastal Restoration Division altei'flora. However,the growthforms of this 1993! specieswere very different in thetwo marshes, The oldersite I wasdominated byshorter S.alterniflora Air temperatureis one metric of effectsof withmore densely packed ramets that did not flower vegetation on surface microenvironmenial duringthe study year. The younger site I was more conditionswhich could affect growing conditions similarto the reference marshes inthat plants were for otherplants or habitatquality for smallanimals. Salt Marsh Developmentan DredgedSediments 228

Cover by S. alrerniflora significantly reduced air ProtectionAgency to the LouisianaEnvironmental temperaturesjust above ground or water! surface ResearchCenter at McNeeseState University. during thesummer, but had no measurableeffect in the winter.Cooler conditions below gra.sscanopy LITERATURE CITED duringthe summer may reduce evaporationof soil moisture.This possibility is supportedby our BraTNEss,M. D, 199]. Zonationof Sparrinaparens qualitativefield observationsthat soil surfacesin and Sparrinaalrerniflora in a New England grassedpatches tended to bc less "dry and cracked" salt marsh.Ecology 72:138-148. duringthe summercompared to areasirnrnediately BEarNESS, M, D, AND A, M. ELi.isON. 1987. outside the grass patch. Other direct or indirect Determinantsof patternin a NewEngland salt effectsof reducedbelow-canopy temperatures could marshplant community.Ecological Mono- occur on seed/seedling viability, surface soil grapfrs57: I 29-147, salinities, invertebrate survival and foraging Gi EASON, M. L 1980. Influence Of tidal inunda- behavior,etc. In s udicsof restored marsh function, tiOn On internaloxygen supplyof Sparrirra moreemphasis should be p! aced on ascertainingthe alrerniflora and Sparrina parens. Ph.D. effectsof dominantvegetati ve cover on the physical Dissertation. Univ. of Virginia, Charlottesville, factors such as sediment organic matter and VA. rtucroclimatefactors affecting conditionsfor other KING,G. M., M. J. Kr i in, R. G, WIEoEitT,AND A, G. plartt speciesand marsh fauna, CHALMERs. 1982. Relation of soil water move m ent and s u 1fide concentraii on to While our study has shown that marsh Sparrinaaberniflora production in a Georgia vegetationdevelops on thesedredge sediment flats, sal t inarsh, Science 218:61-63. questionsremain about differences in S.alterniflora LOEISiANADEPARTMENT oF NATURAI RESOL'RCES, size andflowering among marshes,patterns of COASTALRESTOEATION DivisiON. 1993, Sabine succession,effects of and effects on! sediment TerracingProject Annual MonitoringRepon. parameters,long-term sediment ContpactIOrl rates, Departmentof Natural ResourcesProject No. fauttatuse, levels of export of plantproduction. and 4351089, the geneticvariability in the dominant species.In MENDELSSOHN',I. A., K. A. McKEE, ANDW. H. addition,further evaluatio~ of the amount and value PATnicic,1982. Oxygen deficiency in Sparrina of marshedge vs. interiorin termsof productivity alrerniflora roots: metabolicadaptation to and animalhabitat is neededfor largeareas such as anoxia. Science 214:439-441. the sites at Sabine National Wildlife Refuge, MrrscH, w. J. ANDJ. G. GossELINit.1986. wetlands, Furtherwork is neededbefore declaring these Van NostrandRemhold, Inc. New York, NY. marshescomplete "successes" in terms of the 539 pp. restorationor creationof fully-functioningsalt Nixox, S. W. I982. The ecologyof NewEngland marsh habitat, high saltrnarshes: a community profile,U.S, Dept,of the Interior, Washington,D.C. ACKNOWLEDGMENTS PONNAMIEauMA,F, N. 1972. ThC chCmiStry Of submergedsoils. Advances in Agronomy Weappreciate access to the studysites and 24:29-95. logisticalsupport by staff of the SabineNational WildlifeRefuge SNWR!. Roy Walters of SNWR provideddata on salinitiesand water temperature from theHog IslandGully canal station.Field assistancewas provided by G. Coto,D.J, Devlin, R. Lnwenfeld,M. Salter,and S, Travis. D,J, Devlin, E. Turner,and anonymousreviewers provided helpfulcomments on the paper,This study was fundedby a grantfrom the U.S. Environmental Low-Cost Wetllnd Restoration and Creation Projects for Coastal Louisiana

R. EUGENE TURNER CoastalFcoiogy Instiritte and Departttrenrof Oceanographyand CoastalScience.s Lour'sana State Uni versiry BatonRouge, Lost tstutart 70&3 USA TEL 225-388-6454 FAX 225-388-6326 email; eurtsrne 0 lsts. edtt

ABSTRACIt Wetlandrestoration and crestlion efforts are subject to differenteconomies of scale,lncludhrg scales of ecohrgical propor5ena An inverseeconomy of scaleapparently operates in thevarious smail and large wetla nd restorationprojects possible or implementedin coastal keuhlana.The S ha' gained averntge $12Jl thousand ha' created+restored!isinversely related to project cast,for projectsranging from $08 to $100 million. There is a 15-fold increasein $ ha' gainedas preject slee increases by a factorof ten. Thereappear to begeneric economies of scaleinherent to simlhtrenvironntental management approaches that represent a comprom- isee of at leastthree attributes:attempts to controlecosystem behavior predictability aad tasehecosystem complexity, and incompleteecosystem knowledge. Smail-scale projects are describedthat are very coateffective, including terracing, backfilling, restorstktn of abandoned agriculturalhtnds built br wetlandscirca 191 5k small crevassesand spoil bankmanagement. Thesmallest projects $2+NO eacb! create land at slowrates @ 5 ha yr'! andtend to be very costeffective $20 to MINIha'!. Thereappear to bemore than enough sites tn applythese small projectsat a rate that wotrldetiuni the anticipated restorationand creationrates resulting frtirmthe current CWPPRA program. Kr ywords: weiland creation, resutration, economy of scale,river diversion, Louisiana, coastal zone.

jtstroduetion of the patternsin yield and agricutturalland cul- tivated to extendto otherendeavors in government, Effective wetland restoration efforts in coastal scientific research, and even household economies. Louisianashould be not only long-bsting decades!. Becoming aware of these economies of scale would but also economically'efficient, Economistshave seem to be a prudent aspectof both financial and longrecognized that there is aneconomy of scale political rnanagernentin anera of limiled resources operatingon agricultural lands, An English manand for wetl and restoration. Frenchman,almost simultaneously Sir James Steuartand Turgot ]761; in SchumpeterI9541 There are a variety of wetland creation, re- observedthat as additionalagricultural land was habilitation and restoration methods possible in newlycultivated, thatthe yields perarea eventually coastal Louisiana whose scale varies widely. Most decreased, presumably because thc new farm land of the Coastal Wetland Planing, Protection and wasof moremarginal value. Turgot,in particular, Restoration Act CWPPRA! projects are indi- recognizedan interval of increasingreturns before vidually large. For example,the 1995review of 4 the interval of decreasingreturns. Subsequent years of projects shows that the averageproject observationshave shown these general observations createdor restored203 haat an averagecost of 52. r million, or Sl2.8 thousandha" for thc avcr~g of From the S>mtros!umRcccat Rcscan tr irr CoasratLcrrrtsiatta: individual projects, The averagecost/area of all NataralSysrcrrr Faactr'orr and Rcspotrsc to ftrarrarr ttrflrtcncc. projects was $28.6 thousandha ' of restored or itozas,L,P.. J.A. trlytnan,C.E Protitt, .'V.N.Rabalars. D J. Recit.atNt R.E. Trtrtrer I artrrrrrs!. l 99rr publrs~ by Lour siana created land Anon 1995!. There are also sevetai SeaGrant College Prop.am. types of smaller projects whose costs and results

229 230 R. E. Turner

are either well-docurncntcd or that have substantial projects and water diversionsrange from $12 estimatesto justify their implementation,These thousand to $76 million, the percentage river latter projects are nrtt common in south Louisiana, diverted variesfrom 0,06% to 19%, and the flow is tnostlyfor what seemt tomc! to be for sociological directedthrough a channelfrom 30 to 900 rn wide. or political reasons. They arc relativelysrnaHer The proposedor actualland gain ratesfrom these projectsthan those funded by CWPPRA becauseof crevassesrange from 5 to morethan 600 ha yr'. their individual cost or impact area. These srnaHer projectsare briefly reviewedhere andthen com- Land gain increaseswith project size or the paredto ourexperience with larger projectsin terms amountof water diverted, but not as fast as project of the economyof scale. costs increase. There is, therefore, an inverse relationshipbetween the costof landgained and the sizeof thecrc vassc Fig. 3!. Notethat updated cost Examplesof Low-Cost Projects estimates are included for the West Bank, Davis Pond and Caenarvon Diversion that were not Small Crevasses based on a summary in availableat the time of the paperby Turnerand Beyer et aL l997 and Turner and Boyerl997! Boyer997!. Neither large nor smallcrevasses, of course,can be built everywhere, Constructing "artificial" crevasses,or cuts in the naturallevee of the Mississippi River, has been attemptedby the US Fish and Wildlife Servicein Agricultural ImposrndmeatRestoration the Delta National Wiidlife Refuge DNWR! to based ort Turner aad Neill i984 and build landwithin the Refuge Fig, 1!. The newly Trepagnier et ai. 1995! constructedcrevasses create emergent wetlands after two yearsof subaerialgrowth at about4-5 ha yr' Impoundments occur throughout south Pig. 2!. The presenttotal cost is <$200 ha' after Louisianawetlands Fig. 1!. Somewere built as six yearsand wiH decline to about $54 ha ' as wildlife refuges and others were formed by additional land builds. There is also documentation interlockingcanal spoil banks. The oldestare of the growthrate of largernatural crevasses, and probablythose built to "reclaim' land for agri- of the costsand anticipatedland gains restoration culture. Theseimpoundments are visible reminders or creation! from currently funded CWPPRA of the era,foHowing passage of the SwampLand projects Anon 1995!. The rangeof theindividual Acts of the last century, when publicly owned swamplandswere sold to individualsin transactions whichoften "lacked the characteristics of pristi.ne 60 honesty" Harrison and KoHmorgen1947!. Some areaswere promoted as having a veryhigh potential 50 for profitthrough agriculture.

40 Ten failed or abandonedcoastal agricultural impoundments circa 1915! were examinedby ra 30 Trepagnieret al. 995! to detemunerecent wetland O restorationrates in former agricultural impound- 20 o /+ mentsthat had failed. Aerial photographyfrom 10 between 1978 and 1988 u as used to determine the percentwetland area and levee length and changes betweenmapping intervals. Onesite gained wetland 0 2 4 6 8 $0 12 between19?8 and 1988,and four sites gained Years wetland between 19&5 and 1988. Wetland area in ling.2- Thegrowth of landin newlycreated crevasses the remainingsites were either stableor declining in theMississippi River delta from Boyeret al, 19971. during the study period. The average wetland Low Coast Restoration 231

Fsg.l. Photographsofvarious small resu!ration projects orpotent!al sites in south 1.ouisiana. Topleft: A recent]y- buihc«>'asse i 1990!. Bottom Left: A networkofspo>1 banks resulting from dredged canals. T !pright: the square !nthe center isthc Delta Farms open water area, which formed from a collapsedlevee during a hurricanein 19'. Tbearea !mmediately tothe north was dra!ned and farmed in 1915.but N astreshwater mars h on1930s aerial photos-M!ddle right: an aerial photograph ofthe Sabme terracing pro>ecn Bottom right: a groundview of a recentlycreated terrace <+s months! !nthc Sabine terracing project. 232 R. E, Turner changerates for al! areas r,urged tron> -A.H to + Vegetationforms initially on the exposedspoil 2,~c yr ' from I,q78 to ! egg. Although pan. of material. Laterthe vegetationextends into thewater. this varialbility in rates ii duc to rniipping inter- but at presently undetermined rates. The cost/ha prctatrrr'niliydfolr>glc factors urliqiic to eachsite gainedof thii type of project is in the neighborhood v erealso intportant. The reiults fr<»na multiple of $3.NN to $8,000fha after 5-! 0 years. regressionmodel indicitted that restoration ii inverselyrelated to impoundtncnt size and directly relatedto leveereduction, Trepagnieret al. 995! Spoil Bank Managementfbased un trsed thi» itatistical nrodel tu eitrrnatc thc costi of material in 'orner et aL 1994a! removing the lcvees and the potential land gaini. Levee. retnoval will enhance wetland restoration Thereis virtually no areaof the Louisianacoast rates at a very favorable cost <$1 ha '! and be that is without a dredgedcanal or channel nearby. suitainabie with little additional management, The These canals are built with various dredging recoveryrate, however, was estimated to be «bout methods to facilitate navigation, belowground l~~:yr ' if'10% of the levee were renrovcd. A rather mineral recovery, pipeline construction,. and clearexanrplc of thc pusiibilitiesare shown in thc trapping. Most canali, however,were constructed areanorth of the open water body known ai 1.!efta to servicethe oil and gasindustry, especially tu float Farms, in the Barataria Basin. The open water in drilling equipment. They arc occasionally re- forrnedlfrom a fornter agricultural impoundment dredged or 'swept'! when they fill in ln an aerial whose leveei collapsed in !960. The area on the view of the marsh, canals appear as straight lines northern border was drained and farmccl in 1 915, with a parallel man-made levee on either side but wa.s a.bandoned sornetirne thereafter, perhaps Fig. 1!. Somecanals are isolated,and othersare after a. hurricane. Thc land manager.C. Breaux, found in dense networks. The linear features of recall.sdragging a tractorout of the marshwhen he canals are in sharp contrast to the anatomising, v asa child pers.comm.!, Aerial photograph»show natural channels that form a dendritic drainage that it was mostly wetland by the 1930sbefore the pattern with naturally low levees. intracoastalWaterway v asbuilt. and it retnainsthat v ay trxiay. Thc material removed to create the canal is depositednearby in a continuous linc of dredged spoilmaterial. This tnan-madelevee, called a spoil Terracing basedon information in bank. usual!y hasdifferent. vegetation {c.g., shrubs} CWPFRA Planning docutnents! from the surrounding wetland. Spoil bank leveei havean averageli fetime of lessrhan 50 years{ Monte Terracing ii a wetland restoration practice in !978!. Gas right-of-way pipclincs may not have south Louisianathat involves pihng matcrialon a spoil banksbecause they are often fr lied in,wn.h the shallowand submerged surface to form an exposed previously dredgedmaterial immediately after the surfaceof dredgedmaterial. This new surface is pipeline is laid {backfilling!: theie 'backfilled not like atypical spoil bank dredgedfor navigation, canals frequently re-vegetateenough to make the which is continuous.in two parallel line», and rises new piant cover indistinguishablefrom the abovenormal high tidei, These terracesare more surroundingmarsh. The aggregatelength of these like a series of disconnected ridges arranged canalsand spoil banks i.i in the rangeof 9 to 1 LXK> perpendicularly to each other, whose surface is km and ] gto 2, XX!k.m, reipectively. trooded by high tides Fig. 1}. The natural Resources Conservation Service has constructed The cumulative impact oi conitructing many several of these in the Chenier Plain, and with iorne small individual canals and lcveei is to impound success.The purpose of the raisedbanks, or terraces, marshes, often unintentionally, and r:auie land loss. ii to dampen the waie energy to allow plants tu Spoil bankschange the flow of water into and out takeroot and suspendedmatter tu fall out in greater of the marsh t'Swenion and Turner I 'N7!, causeopen amounti, and to protect the adjacent shoreline. water pond~to fornr nearby Turner and Rao i 99 tu Low Coast Restoration 233

and are spatia!lyand tempora!!yrelated to the water habitat is, therefore, the general goal of conversionof wct!andhabitats to openwater e.g., backfilling.Canal backftlling i» generallyintended Bass and Turner 1997;Turner ! 997!. to directlyachieve four objectives:

g impactsresult from the: ! longer !! To fill in the canal., wetlanddrying cycles. even in setni-impounded 2! Toreestablish emergent marsh vegetation wetlands,as a consequenceof altered water move- in the cana! tncntsinto and out of the wetland. Thc lengthened 3! To eliminatethe spoil bank, and, drying periodspromote soil oxidation and subse- quentsoil shrinkage; ! floodingevents that may 4! Toreestablish emergent marsh lengthenbehind spoil banks, presumably as a vegetationon thc spoil bank. consequenceofwaterbeing trapped behind the spoil bankonce water enters overland during very high Backfi!lingbegan to bc requiredin !979 as a tides Whenwetland flooding increases enough to conditionfor issuing a permitto dredge some canals. seriouslywaterlog soils and change soil cheinistry, Canalsmay be backfillcdafter thedrilling site is plantsmay becomestressed to thepoint where abandoned on-sitc mitigation!, or, anothercanal growthreduction or evendie-back occurs e.g., may be backfilled off-site mitigation! to obtaina Mendelssohnet al. 1981;McKce and Mendelssohn permitfor dredginga newcana! or otheractivity. 1989!;! lowersedimentation rates behind spoil The spoilbank is leve!edas near to marshelevation banksin any wetlandtype, because of thereduced aspossible, and the spoiI bankmaterial that is moved frequencyand depth of tidalinundation; ! ln isused to fill inthc existing canal. The fill is placed addition,the spoil banks consolidate theunderlying evenly over the bottom of the canal, soi!s, Subsurface water movement isthus decreased, bothbecause of thereduced cross-sectional area and Whi!ethe backfilledcanal and adjacent berm the reducedpermeability of materialbeneath the westudied was not restored to a completelynatural levee Turner 1987!, condition,its status wasintermediate in many respectsto thatat a naturalbayou andan unfilled Thereis nosystetnatic manageinent plan ta canal Turner et al, 1994b!, restorespoil banks to wetlands byremoving them. We useda hydrolagicmodel and fie!d data to Thereduction in c!cvationof the spoil bankat estimatethat removing about ! 0%of the spoil bank thc backfilledcanal resulted in greaterwater wouldrestore hydrol ogic fl ows if thespoi/ retttovat movement in and ou of the marsh, redticed the werestrategicaliv pktc

10 EO~

-10 0 4 10 j'

3 10 EOIO 10 7 10 10 PejectCost $ j ProjectCost $! Fig,4, The re!atiottship betweenthe cost per ha land created orrestored andthe project cost for al! CWPPRA ptojectsinsouth Louisiana. ittcludlngdiversions, Left:$ha' cost vs. project cost.Right; Average AnnualCost perAverage AnnualHabitat Unit AAC/AAH 1 vs.! project c ost. Thedata arefrom Anon 995!.

1 00,000

10,000

1,000 a C

100 w

10

I 00 1 000 10,000 1 00,000 Project Size $1000s! "ig-5-The relationship betweenvarious small and large restoration projectsincoastal Louisiana implemented and uu-imp!ementedprojectcostsare averaged!. Therelationship betweeri restorationrcreation projectsizeand hecoat P hawetland restored orcreated. Theliterature sources arefrom: CWPPRA Coastal Wetland Planruog, Protee

have inherent requirements that dictate, to some ACKNOWLEDGMENTS extent,where they can he implemented. These constraintsmay require adjustments in management This study was originally funded by two US efforts to achieve optimum uses of the available EPA grantsthrough the Gu/f of Mexico Program fundingand ecological resources, It is worthnoting, andEPA Region6, Dallas, Texas. We thankthe however,that thc vast majority of CWPPRAprojects National Marine Fisheries Service o%c e at LSU for are large projects. The first 4 years of the CWPPRA repeateduse of theiraerial photographs, progtamhadonly 1 of63 projectsthat costless than $5OO,GOO. LITERATURE CITED lt is worth askingif there is enougharea to use there smaller projects. Only preliminary numbers ANoN. 1995. CoastalWetlands Planning, Protec- are available,but they suggestthat smallprojects tion, andRestoration Act: Sutnmaryof Priority are a viable option for restoration efforts. The Lists 1-4. CWPPRA Report Series No. l. CWPPRAprogram has projects in placeto attempt LouisianaDepartment of Natural Resources. to create and restore wetlands at a rate of 479 ha BatonRouge, Louisiana. yr' Anon 1995!. A stnall crevasse builds at 4.76 BASS,A, ANuR. E. TURN>R.1997. RelatiOnships ha yr' Boyeret al1997!, so it would take only betweensalt marsh loss and dredged canals in I OOsmall crevasses,at $48/ha, to create what costs three south Louisiana estuaries. Journal nf the CWPPRAprogram $28,000 h 'a, There are a Coastal Research 13:895-903. few hundred locations available to build crevasses Bxuxa~NN, R. H. AND R. E. TURNER. 1990, Direct Boycr and Turner 1995!, but land owner iinpacts of outer continentalshelf activitieson permissionis required something that may not be wetland loss in the central Gulf of Mexico, forthcoming, The open water in abandoned Environmerital Geology and Water Resources agriculturalimpoundments could be recoveredat 15: I 89-198. 10ha/yr, Trepagnier et al., 1995!,at a costof about $1ha'. Thereare at least 43 unpoundments,perhaps BovER, M. E. wNo R. E. Ttrxma. 1995 Data on 20of whichcould be restoredwith permission.The crevassesplays in the Delta National Wildlife averagesize is 2,352 ha. The potentialrestoration Refuge,Mississippi RiverDelta. Final Report to the United States Environmental Protection atea is, assuming50% recovery, 23,520 ha, at 50 years,or 470 ha yr', equivalentto the CWPPRA Agency,Dallas, Texas. landgain at .01% of theper hacost. The greatest BDYER, M., J. HARRts xNo R. E. Tuxr RR. 1997. opportunity to implement low cost restoration Constructedcrevasses and land gain in the projectshas to dowith restoring oil andgas recovery MississippiRiver delta Journal of Restoration canalsand reversing their negative impacts on Ecology 5:85-92, wetlandhydrology. There are tens of thousandsof HxatuSON, R, WxNn W. M. Komwoxoan. 1947. spoil banksand canalsthat could be backfilled, Pastand prospectivedrainage reclatnations in equaling 80,426 ha in 1978 Bauman and Turner coastalmarshlands of the MississippiRiver 1990!,There is a muchlarger area of impoundments Delta, Journal of Land Public Utility andsemi-itnpoundments that couM bemanaged, an Economics 23: 297-320. areathat perhaps exceeds 500,000 ha. It wouldbe MCKaa, K, L, axo L A. MRNoxt.SSOitN 1989. necessaryto convertonly 0.15 % yr' of the canals, Response of a freshwater marsh plant spoil banks,semi-impoundments and itnpound- conununity to increasedsalinity and increased mentsto wetlandseach year at lessthan $1000 waterlevel. AquaticBotany 34: 301-316. ha'!,to matchthe objectives of theprojected results MENDELSSOHN,1. A., K, L. MCKai= ANo W. H. of currentCWPPRA projects. PATRtctc.JR. 1981, Oxygen deficiency in SparrirtaalrerntIlora roots: metabolicadapta- tion to anoxia. Science 214;439-441. Low Coast Restoration 237 MOxTa,J, A 1978. Theimpact Of petrOleutn dredgingon Louisiana's- coastal landscape: A plantbiogeographical analysis and resource assessmentof spoil. Ph,D. Dissertation, LouisianaState University, Louisiana. NEtts.,C. ANDR. E. TURivEtt. 1987. BaCkfilling canalstomitigate wetland dredging inLouisiana coastalmarshes. Environmental Managemerrt 11:823-836. ScHUMpETER,I, A. 1954. History of Economic Analysis.Oxford University Press, New York. Svrpvsoz,E. lvl. ANn R. E.Titttxatt. 1987. Spoil banks;Effects on a coastalmarsh water level regime.Estuarine, Crrasral and Shelf Science 24:599-609. TttEPAGvttP3t,C. M., iv!, A. K Xths,A Yl!R. E. TL'RNER. 1995.Evaluation of wetland gain and loss of abandonedagricultural impoundments inSouth Louisianafram 1978to 1988. Resroratiorr Ecology3/4;299-303. Tt ttrtEa,R. E. 1987,Relationships between canal and levee densityand coastalland lossin Louisiana.United States Fi.ch and Wildlife ServiceBiological Report 854!. TURv'Ht,R. E, ANDC. NrttL. 1984.Revisiting impoundedwetlands after 70 years, p. 309-322. In R.3.Varnell ed.!, Water Quality and Wetland ManagementConference Proceedings New Orleans,LA., August 4-5, 1983 spon sors: LouisianaEnvirottmenta1 Professionals Asso- ciation!, Tuttrtatt,R, E. htvo!Vl. E. BovLR. 1997. Mississippi Riverdiversions, coastal wetland restoration/ creationand an economy of scale, Ecological Errgineering8:117-128. TUttstt:tt,R, E AxoY. S. Rho. 1990. Relationships betweenwetland fragtnentationand recent hydrologicchanges in a deltaiccoast, Estuaries 13;272-28 l. Ttjtt~~,R. EE.M. SwpvSO~,hr~ J.M. LPL,1994a. A rationale for coastal wetlandrestoration throughspoil bank rnanagernent in Louisiana. Envirnnrrrental hfanagerrtent 18, 271-282. TLtc~xtt,R. E, 3. M Li E. A~OC. NatLL. 1994b, Backtilling canalsas a wetland restoration technique in coastal Louisiana. Wetlands Ecologyand Management3:63-78.