Upload
william-j
View
215
Download
2
Embed Size (px)
Citation preview
J. Great Lakes Res. 18(4):552-570Internat. Assoc. Great Lakes Res., 1992
COMBINING ECOSYSTEM AND LANDSCAPE APPROACHESTO GREAT LAKES WETLANDS
William J. Mitsch l
School of Natural Resources and Environmental Science ProgramThe Ohio State University
Columbus, Ohio 43210-1085
ABSTRACT. Great Lakes wetlands can be studied systematically and for their importance in thelandscape at different scales. A systems-oriented study of the wetlands oj southwestern Lake Eriealong Ohio's shoreline has been organized at two levels ojhierarchy. At the ecosystem level, a specificwetland model was developed from data collected on hydrology, nutrient budgets, and aquaticmetabolism. The model, calibrated from the field data, was used to predict phosphorus retentionrates and to compare them with results from empirical models and field studies. At the landscapescale, synoptic surveys of diked (hydrologically isolated with impoundments) and undiked (naturalhydrology) wetlands for hydroperiods, water quality, sediment chemistry, and vegetation biomassand species illustrated several differences between these two types of wetlands. Combining datacollected from these surveys with data collected from remote platforms will lead to the developmentof spatial dynamic models for the shoreline to deal with landscape-level questions on the management of these shoreline wetlands.INDEX WORDS: Ecological model, hierarchy, Lake Erie, phosphorus, Sandusky Bay, spatialmodel, water quality, wetland vegetation, wetland sediments, freshwater marshes.
INTRODUCTION
Coastal wetlands on the shoreline of the Laurentian Great Lakes are unique in the world of wetlands. The wetlands and their associated rivershave been described as parts of estuaries by some(Herdendorf 1987) and rejected as estuaries by others (Schubel and Pritchard 1990). In presettlementtimes, the vegetation of these coastal wetlands expanded and retreated with changing water levelsyet always remained as a buffer between the uplands and the lakes. As shorelines were stabilizedand the land was drained for agriculture and urbandevelopment, these wetlands were mostly destroyed or significantly altered; their buffering capacity was diminished or lost altogether. Many ofthe wetlands that have not been drained for shoreline development have survived because they havebeen impounded as wildlife habitat, thereby isolating them from the Great Lakes and watershedsthat formerly nourished them.
IThe author acknowledges the significant contributions of his graduate students, particularly Brian Reeder and Doreen Robb, and thecollaboration by Mary Roush, Oi-Chul Yi, and Oreg McNelly.
552
Few if any comprehensive systems-level studieshave been carried out on these Great Lakes coastalwetlands (Prince and D'Itri 1985). This is particularly apparent when comparing Great Lakes research with the abundant "ecosystems" literatureavailable on estuarine and coastal salt marshes(e.g., Nixon and Oviatt 1973; Nixon et al. 1976;Hopkinson and Day 1977; Kremer and Nixon1978; Pomeroy and Wiegert 1981; Wiegert et al.1981; Costanza et al. 1988, 1990). This paper summarizes highlights of an ongoing hierarchical studyof a region of wetlands on the southern shore onLake Erie in Ohio. Models of wetlands are beingused to guide field research efforts, to identifygaps in data, to investigate ecosystem behavior,and ultimately to aid in wetland management bypredicting systems-level behavior of coastal wetlands under different management conditions. Ourapproach uses two levels of hierarchy of modelsand field measurements, ranging from an ecosystem level approach that emphasizes ecosystem processes in individual wetlands to landscape approaches that cover large geographical areas.
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 553
Loadings from the Upland Watershed
Coastal Great Lakes wetlands are influenced byrunoff from upstream watersheds which are oftendominated by agricultural use and are thereforesignificant sources of nutrients, pesticides, andsediments to both these wetlands and to Lake Erie(Yaksish et ai. 1982, Baker 1988). For example, the69 km2 primarily agricultural watershed that drainsinto Old Woman Creek wetland on Lake Erie contains relatively high levels of nutrients (Heath1987, Klarer 1988, Klarer and Millie 1989) and wasestimated to discharge 0.5 to 1.0 kg P ha-1y-l to thewetland (Mitsch et ai. 1989a). Many counties in
AN OVERALL MODEL
In any effort to understand and model an ecosystem or region, an understanding of the forcingfunctions or driving variables of that ecosystem isparamount. Great Lakes wetlands have severalforcing functions that, collectively, make themunique ecological systems to study and manage(Fig. 1). These include: 1) varying hydrologic, nutrient, and toxic loading from upstream watersheds; 2) shifting shoreline sediments, moved during storm events, which can dramatically changethe hydrologic, chemical, and biological connections between the wetlands and the Great Lakes; 3)water level fluctuations of the Great Lakes, whichvary both seasonally and annually; 4) periodicseiches or "wind tides"; and 5) artificial impoundment construction and water level manipulation byhumans. These forcing functions have one thing incommon - they greatly influence the exchange ofgeologic and biological materials from upstreamwatersheds to and from the Great Lakes.
Great LakesCoastal
Wetlands
Diking,fo4---IManagement
Ohio in the western Lake Erie basin have dischargerates of 0.5 to 2.5 kg P ha-1y-l to the lake (Johnsonet ai. 1978, Novotny 1986). A study group of GreatLake's pollution called PLUARG (Pollution fromLand Use Activities Reference Group) presents arange of 0.1 to 9.1 kg P ha-1y-l (IJC 1980).
Shifting Shoreline Sediments
For most of the year, the general direction of flowthrough open coastal wetlands is from the uplandwatershed, through the wetland, and to the GreatLakes. However, the difference in elevation between a coastal wetland and the lake itself can varywith storm events and short-term Great Lake fluctuations. The difference between the two levels canbe exacerbated when the mouth of the stream between the wetland and the lake is closed due toshifting shoreline sediments, a rather frequent evenin some wetlands (Mitsch and Reeder 1992). Thewetland can remain closed for several months, after which the combination of high water levels inthe wetland and a sudden storm event once againopens the wetland to the Great Lakes.
Lake Water Fluctuations
Wetlands along the Great Lakes are heavily influenced by water level fluctuations of the GreatLakes. For example, water levels for Lake Eriewhich have been recorded over the past 130 yearsnear Cleveland, Ohio (Fig. 2) show a difference ofalmost 150 cm between low and high water levels inthat period. This amplitude is significant enoughto dramatically affect shoreline processes, particularly those in wetlands. The time between high andlow water levels (approximately 10 to 15 years recently) is long enough to affect the structure andfunction of coastal wetlands. In presettlementtimes, high water levels would send the wetlands"inland" (Fig. 3a) while wetlands would extend"lakeward" during low water levels (Fig. 3b), oftenleaving the wetlands of the Great Lakes in a stateof disequilibrium with its water level. At a givenlocation, the wetland will vary from a system dominated by emergent vegetation (during shallow water times) to one that is a planktonic or floatingleaved aquatic system (during high water level).
FIG. 1. Conceptual model showing major jorcingjunctions that affect Great Lakes coastal wetlands.
Seiches
Shorter period water level oscillations due to windaction, called seiches, frequently occur on the GreatLakes. The coastal wetlands along the lakes are subject to water and chemical exchanges from seiches
554
573.50
572.50
571.50
570.50feet
(aboveMSL)
569.50
568.50
567.50
W. J. MITSCH
Lake Erie Water Levels· Cleveland
175
174.5
174
m(aboveMSL)
173.5
173
1860 1870 1880 1890 1900 1910 1920
year
1930 1940 1950 1960 1970 1980
FIG. 2. Water level in elevation above sea level ofLake Erie from 1860 to 1988. Data are from station at Cleveland,Ohio, and are courtesy of K. Baker, Ohio Department of Natural Resources.
in much the same way that coastal salt marshes aresubjected to tides, although these seiches are not asperiodic as semi-diurnal coastal tides. Sager et at.(1985), for example, measured 269 seiche events in 1year on lower Green Bay on Lake Michigan with amean amplitude of 19.3 cm and a mean period of9.9 hours. Herdendorf (1987) reported that longitudinal seiches occurred in Lake Erie approximately40% of the time and most often at 12-14 hour periods. The importance of seiches to the nutrient budgets and biotic communities of Great Lakes wetlands is not well known although it may beanalogous to the role of tides in salt water wetlands,transporting nutrients and detritus into and out ofthe wetland. Impounded wetlands are isolated fromexchanges due to these seiches.
Artificial Dikes and Water Level Management
The fluctuating, less predictable water level of theGreat Lakes has led to a common practice in marsh
management along the lakes' shorelines of constructing artificial impoundments (dikes or levees)and maintaining artificial water levels with pumpsand gates. Levees are constructed around wetlandsand pumps or flap gates are installed to keep waterlevels independent of those of the lakes. Duringlow water times or periods of drought, pumps areused to keep water in these managed wetlands (Fig.3c). Because many of the coastal wetlands alongthe Great Lakes are managed for waterfowl, dikedwetlands are a common type of wetland there, especially along southwestern Lake Erie.
A HIERARCHICAL STUDY OF WETLANDSOF WESTERN LAKE ERIE
Our wetland study area centers on the wetlands inthe region of Sandusky Bay on the southwesternshore of Lake Erie in northern Ohio. The area isapproximately 60 km wide and is bounded on theeast by Old Woman Creek wetland, approximately
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS
HIGH LAKE LEVEL
FIG. 3. Dynamics of wetlands of coastal GreatLakes during a) high lake levels when wetland vegetation moves landward, b) low lake levels whenwetland vegetation moves lakeward, and c)impounded or diked wetland where water level iscontrolled by pumping and levees.
555
5 km east of Huron, Ohio, and on the west by themouths of Sandusky River and Muddy Creek asthey enter Sandusky Bay southwest of Port Clinton,Ohio. We have chosen initially to look at the LakeErie wetlands in this region on two scales of hierarchy (Fig. 4). Intensive studies and measurements ofecosystem productivity, nutrient cycling, and paleolimnology have been carried out at Old WomanCreek wetland. A landscape assessment of hydroperiods, water quality, sediments, and productivity,which includes several other marshes along LakeErie, is included in the study to determine if theintensive studies at Old Woman Creek can be generalized to coastal wetlands along Lake Erie. Detailsof these studies are contained in Mitsch et al.(1989a); Mitsch (1989); Robb (1989); Reeder (1990);Mitsch and Reeder (1991, 1992); and Robb andMitsch (1992).
ECOSYSTEM SCALE
Our primary site for ecosystem-level studies was atOld Woman Creek National Estuarine Research Reserve and State Nature Preserve, a natural LakeErie coastal wetland in Erie County, Ohio. Its hydrology is determined both by Lake Erie water levels and runoff from a 69 km2 watershed. The wetland itself is 56 ha in size and extends about 1 kmsouth of the Lake Erie shoreline (Fig. 4). Depthsmay extend to 3 m in the inlet stream channel, butfor the major portion of its area it is usually lessthan 0.5 m deep. The wetland has an outlet to LakeErie that is often open but which can be closed forextended periods of time by a barrier beach. Seicheson Lake Erie can reverse the normal lake inflow,causing lake water to spill into the wetland. Aquatichabitats within the wetland include open waterplanktonic systems and American lotus (Nelumbo
556 W. J. MITSCH
kllomoter
$802488
01 23" 5mile
LANDSCAPE SCALE
• Hydroperiod• Water Quality• Wetland Sediments• Wetland Vegetation• Remote Sensing• Spatial Model
ECOSYSTEM SCALE
• Hydrology Budget• Aquatic Metabolism• Paleoecology• Ecological Model• Chemical Budgets• Chemical Processes
FIG. 4. Two scales ofhierarchy for study ofLake Eriewetlands in Ohio, including types of studies that werefound to be most appropriate at each level (Mitsch1989). Abbreviations in upper map refer to selected wetlands: WPW = Winous Point West (diked); WPN =Winous Point North (diked); OSCB = Ottawa Shooting Club Big Pond (diked); OSCA = Ottawa ShootingClub Allen Pond (diked); PCK = Pickerel Creek (natural); WLP = Willow Point (natural); BVC = Bay ViewCenter (diked); BVB = Bay View B (diked); PLMB =Plumb Brook (natural); SHM = Sheldon Marsh (natural); OWC = Old Woman Creek (natural).
lutea) beds throughout a significant portion of themarsh.
Ecosystem Simulation Models
Figure 5 shows a model of Old Woman Creek wetland (Mitsch and Reeder 1991) with details of someof the processes in a wetland which contribute toits nutrient retention capability. Plant uptake, bothby plankton and macrophytes, sedimentation, andresuspension are probably the most significantprocesses involved in the wetland retaining and re-
leasing phosphorus. This conceptual model wasdeveloped into a simulation model based onknowledge of the cycling of phosphorus and energy in wetland ecosystems. Unique to this wetlandmodel is the interaction of Lake Erie with the wetland. The model was divided into three submodelsand simulated by the higher level simulation language STELLA™ (Mitsch and Reeder 1991).
A hydrology submodel of the simulation model isdesigned to depict the hydrologic budget of OldWoman Creek wetland with the only state variablefor this submodel being the volume of water in thewetland. Factors affecting the volume of water inthe marsh include rainfall, watershed inflow,evapotranspiration, and exchange with Lake Erie.The availability of inflow data, fairly good data onevaporation, and knowledge of the hydrologic forcing functions in the wetland allowed the development of an accurate hydrologic budget and model.An important part of the hydrodynamics is the timing of an ephemeral barrier beach on the wetlandoutflow.
A primary productivity submodel includes thestate variables macrophyte biomass and planktonbiomass. These are both a function of sunlight, withthe major losses being respiration and sedimentation. The productivity submodel is linked to thehydrology submodel in two ways (Fig. 5). First,plankton are exported to Lake Erie when waterfrom the wetland flows into the lake and the beachis open. This assumption is consistent with the fielddata. Second, macrophytes are assumed to be moreabundant when water levels are lower, and lessabundant when water levels are higher. This is basedon observations of other Lake Erie coastal marshes(Herdendorf 1987). Field measurements of grossprimary productivity (see below) as measured at OldWoman Creek wetland were in general agreementwith model simulations. Using chlorophyll a as acalibration variable, the model and field data werefound to be in agreement (within one standard deviation) 78070 of the time. Given the differences in thetwo field measurements and the variability ofenergy/chlorophyll ratios in aquatic systems (Vollenweider 1974), the model accurately predicted seasonal patterns of productivity and biomass (Mitschand Reeder 1991).
A phosphorus submodel was coupled with boththe hydrology and primary productivity submodelsand is incapable of running simulations withoutinput from these submodels. This submodel utilizes one phosphorus storage in the waters of thewetland and another in the sediments, with linear
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 557
--Energy
-- Phosphorus
Permanent Burialor Phosphorusin Sediments
FIG. 5. Conceptual model of Old Woman Creek wetland along Lake Erie (fromMitsch and Reeder 1991a). This model describes the major pathways in a simulationmodel developed for that wetland.
pathways between the two. The phosphorus submodel includes a sedimentation pathway as defined for shallow lakes by Kamp-Nielsen (1983)and Henderson-Sellers (1984) with an average settling velocity of 0.03 m d-1
• Calibration was doneby varying the resuspension coefficient until themodel predicted phosphorus concentration resultssimilar to field data.
Hydrology, productivity, and phosphorus fielddata enabled model calibration and preliminary estimations to be made of the role of phosphorussedimentation and phosphorus resuspension in theshallow wetland. Simulations show high levels ofsedimentation in the early spring, with resuspension exceeding sedimentation through the remainder of the year (Fig. 6a). This excess of phosphorus resuspension over sedimentation in the
model simulations is surpnsmg at first, but theproductivity estimates throughout the year clearlyillustrate that there is insufficient phosphorus inthe inflow to support the high level of productivityand the generally high phosphorus concentrationsexperienced in the wetland from May throughNovember.
The model also predicts the total phosphorussedimentation rate, including contributions fromplankton and macrophytes (Fig. 6b). Sedimentation rates as high as 40 mg P m-2d- 1 are simulatedfor the spring, while the rate for the remainder ofthe year, when very little allochthonous inflows areexperienced, is around 10 mg P m-2d-1
• The simulated rates of sedimentation and resuspensiontranslate to a total net sedimentation of 0.8 g P m-2
for the 9-month study period. This contrasts to an
558 W. J. MITSCH
40000
30000
g-P/day20000
I resuspension2 sedimentation
10000
0.061 130 198 267 335
Time, days
IMar Apr I May Jun Jul I I
Aug Sep Oct Nov
0.0400
0.0300
g-P/m2-day
0.0200
Total Sedimentation Rate
0.01000
61
I Mar
130
Apr I May Jun
198Time
Jul Aug
267
Sep Oct Nov
335
FIG. 6. Calibration simulations from Old Woman Creek wetland model that demonstrate approximate rates and seasonal patterns ofsedimentation and resuspensionfrom March through November (Mitsch and Reeder 1991). Total sedimentation rateincludes contributions from planktonic system.
estimated annual retention of 5-7 g P m-2 predictedby us a few years earlier using a simple empiricalmodel (Mitsch et af. 1989). Because the model isbased on the year 1988 which had a significantdrought, we can expect this net sedimentation rateto be well below average. It is not unreasonable tosuggest that less than 20 percent of the normalsedimentation occurred in the wetland during ourcalibration year of extreme drought and as a resultof having a simulation for only nine of the twelvemonths in the year. Subsequent simulations
(Mitsch and Reeder 1991) show that higher inflowsfor the same 9-month period lead to proportionately higher net retention of phosphorus, approximately 1.3 to 3.3 g P m-2
, respectively, for normaland wet years.
Ecosystem Metabolism
Diurnal patterns of dissolved oxygen can give indications of systems-level parameters of wetlandsthat are useful in assessing biological metabolism
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 559
9 12nidniglt
6
Ststlon 8, Old WomsnCr..k WstlsndJuly 11-12, 1988
9 12 3 6 9 12 3midnight noon
Time, hour
3 6 9 12 3 6 9 12 3 6 9midnight noon
4
2
6
.2
.4
.6
26
25
24
12
14
-.6 t--.----r-.--.---r-.--.----.-.--.----.-+12 3 6
noon
10D.O.,mgA 6
-.4
3 6 9 12 3 6 9 12 3 6 931 midnight noon
30
Temp, 26
·C 27
29
16..-- ......
D.O.
Rate of 0 +----+-----rl-"-<--~____>O',.__"".....,\____+Change,
ghn~hr -.2
30 kcal m-2 d-1 in July yields an efficiency of 0.4 %.The average gross primary productivity of 8 kcalm-2 d-1 in October results in an efficiency of 0.3 ltfo.These productivities are significant for planktonicsystems and are comparable to those of productivewetlands in flow-through conditions (Mitsch andGosselink 1986). P/R ratios calculated from Julydata suggest an excess of production over respiration at six of the eight sites although, interestingly,there was greater respiration than production atthe two sites in the upstream reaches of the wetland. All sites except one in October have P IRratios less than 1.0 as solar energy and tempera-
FIG. 7. Typical summer diurnal patterns of dissolvedoxygen, temperature, and rate of change of dissolvedoxygen for Old Woman Creek wetland. Note that dissolved oxygen ranges from 3 to 15 mg L-1
•
and in calibrating ecosystem models. The rise andfall of dissolved oxygen in an aquatic system reflect that system's metabolism. Diel patterns of oxygen have been used to determine the overall primary productivity and respiration of estuarine,lacustrine, and riverine systems (e.g., Odum andHoskin 1958, Mitsch and Kaltenborn 1980, Meyerand Edwards 1990). Shallow plankton-dominatedwetlands such as the Old Woman Creek wetlandnear Lake Erie are ideal systems in which to measure the diurnal patterns of oxygen and to usethose patterns to estimate productivity. The shallow nature of the wetland limits the euphotic zoneto a narrow depth of high chlorophyll and dramatic oxygen swings. The warm water temperatures in the summer season further enhance biochemical activity. Furthermore, the calm waters ofthe shallow wetland usually do not have high ratesof oxygen diffusion. Measurement of the diurnalpatterns of water chemistry parameters such as dissolved oxygen and temperature is also important topredict redox conditions and the habitat value ofsuch wetlands for aquatic organisms.
Dissolved oxygen patterns were measured onseveral occasions with Winkler titration and metersin Old Woman Creek wetland (Mitsch 1989,Reeder 1990). Gross primary productivity and respiration are calculated from rates of change of thedissolved oxygen, with nighttime readings givingan estimate of the hourly rate of respiration anddaytime changes reflecting both gross primary productivity and daytime respiration. For example,July 1988 dissolved oxygen data display a dramaticchange in oxygen over a 24-hour period (Fig. 7).Readings ranged from 2 mg L-l at 6 AM to peaksof 12 to 15 mg L-l at dusk. Water temperatureswere quite warm at 24 to 29°C. In contrast, thedissolved oxygen range in Old Woman Creek inearly October was less dramatic, ranging from 6 to10 mg L-l with a temperature range of 6 to 14°C.Water during July was clearly supersaturated inlate afternoon with dissolved oxygen and undersaturated at dawn; numbers remained much closer tosaturation during the October readings.
Old Woman Creek wetland, with the higher thannormal depths prevalent during this study year, isprimarily a plankton-dominated system. Gross primary productivity calculations, based on the Julydata, result in measures of 15 to 57 kcal m-2 d-1 oran average of 30 kcal m-2 d-1 for all sites. The October data result in productivity calculations of 2 to19 kcal m-2 d-1 with an average of 8 kcal m-2 d- l forall sites. An average gross primary productivity of
560 w. J. MITSCH
tures decrease rapidly in October and the systemshifts to catabolism and lower productivities.
Hydrology Budgets
Old Woman Creek wetland has complex hydrologic conditions typical of natural Great Lakeswetlands still connected to their lake. For example,the water level of the wetland reached a high of4.56 m above datum (approximately 1 m deep) on8 May 1988 before a 1.4 cm storm led to a breakthrough of the barrier beach and a water level dropto 55 cm in 2 days. The barrier beach was open formost of March, half of April, and for a short period during May. After those openings, the beachclosed for the remainder of the measurement period. It is normal for the wetland to be opened toLake Erie during the spring and then closed duringthe summer. The wetland was open to Lake Eriefor a total of 50 days, or 23 percent of the studyperiod. An estimate of the daily water budget forthat study period, calculated from the water levelsand the water budget equation, demonstrated thatsurface inflow from the watershed averaged 15,200m3 d-1 while Lake Erie provided 3,500 m3 d- I (Fig.8). The flow from Lake Erie represents 18070 of thewater flow into the wetland. As expected, evapo-
transpiration was higher than precipitation (by80%) because of the summer drought. The watervolume in the wetland actually increased slightlyduring the study period (March-November 1988)by 700 m3 d-I
, even with the drought conditionsbecause of the closed barrier beach between it andLake Erie and because of possible groundwaterinputs.
Chemical Budgets
Table 1 contrasts rates of phosphorus retention inOld Woman Creek from empirical model calculations, from field data collected in 1988 (Reeder1990, Mitsch and Reeder 1992), and from the simulation model (Mitsch and Reeder 1991). An empirical loading retention model of Richardson andNichols (1985) predicted 8-19 mg P m-2d-1 beingpermanently buried in the wetland sediments understandard phosphorus loading rates from Lake Eriewatersheds estimated from Johnson et af. (1978),IJC (1980), and Novotny (1986). This method ofestimating phosphorus retention does not take intoaccount any of the interactions between Lake Erieand the wetland nor does it consider any intrasystern interactions. A mass balance predicted fromfield data (Reeder 1990, Mitsch and Reeder 1992)
Precipitation
Units are m3/day x 1,000
Evapotranspiration
1.0 1.8
Lake Erie~17.2
Outflow toLake Erie
Inflow fromLake Erie
3.5
,IrA A A A A A A A A A A A A "",A"", A A A A A A AA AAA A #to A
A:A Old Woman Creek :A:A:A: Wetland A:< ....A AAA A A A
.... ""AI. """"'''"A
:<A lJ.V/lJ.t = +0.7 A>:A A A A
A A A AA A A A A A A A A A A A A A A A A A A A A
A A A A A A A A A A A A A A A A A A A A A AA A A A A A A A A A A A A A A A A A A A A
A A A A A A A A A A A A A A A A A A A A A A
15.2
Surface Inflow
FIG. 8. Approximate water budget for Old Woman Creek for March through November, 1988 (from Mitsch andReeder 1992).
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 561
TABLE 1. Comparison of nutrient retention capabilities of Old Woman Creek wetland by different measurements.Numbers except % retention are in mg P1m2-day.
Method Inflow Outflow Retention 070 Retention Source
Empirical Model v. I 36-63 14-19 30-39 Mitsch et al. 1989aEmpirical Model v. II 17-33 8-13 39-47 Mitsch and Reeder 1991Field Data 2.2 1.4 0.8 36 Reeder 1990, Mitsch
and Reeder 1992Sediment Core 22 Reeder 1990Simulation Modela
Dry Year 32.3 20.6-26.8 5.5-11.7 17-36 Mitsch and Reeder 1991Normal Year 45.5 23.6-33.1 12.4-21.9 27-48Wet Year 83.7 39.9-63.8 19.9-43.8 24-52
afor 9-month period only; high % retention in range is with high lake level, low % retention is with normal lake level
has a much lower loading rate primarily because thedata reflect drought year conditions and do not include an entire calendar year. Phosphorus retentionfrom the field data was estimated to be 36070 of theinflow, nearly the same retention percentage as predicted by the empirical models. The field dataresults indicate that the contribution of Lake Erie tophosphorus loading to the wetland is minimal(0.4% of watershed inflow) as assumed in themodels. Actual loading rates and retention may behigher than either the empirical model or field datapredict, because data from one recent sediment coreshow that an average of 22 mg P m-2d-1 has beendeposited over the past 180 years (Reeder 1990).However, the sediment coring site is in the area ofthe marsh with the highest sedimentation rate (Buchanan 1982), so the sedimentation data probablyrepresents a region of maximum retention.
The simulation model described above takes intoaccount the effects of hydrology on loading rate,the amount of transformation occurring due to thebiota, and the sedimentation and resuspension estimated from model calibration (Table 1). The simulation model predicts that 5.5 mg P m-2d-1 or lessthan 20% of the phosphorus inflow is permanentlyretained in the wetland - more than the field estimates suggest, but less than the empirical modeland sediment core predict. As expected, the primary producers transform some bioavailable phosphorus into non-bioavailable forms, but this contribution to the sediments is minimal in the model.Other simulations predicted that as much as 44 mgP m-2d-1 could be retained in the marsh, especiallyunder high water level conditions (Reeder 1990,Mitsch and Reeder 1991). The model estimates arereasonable considering the variability of the dataused for comparison, and probably represent the
best available estimate on phosphorus retention inthis Great Lakes coastal wetland.
LANDSCAPE-SCALE WETLAND STUDIES
The landscape-scale area of Lake Erie shoreline istoo large to address with ecosystem-level measurements and ecosystem models, but it is closer to thescale that is required to determine the importanceof coastal wetlands on Lake Erie. We initiallychose 11 wetland sites including Old Woman Creekfor synoptic studies of hydroperiod, water quality,sediments, and vegetation and for the applicationof remote sensing, geographic information systems, and spatial modeling (Fig. 4). Several ofthese wetlands are diked to maintain artificial water levels, primarily to attract waterfowl. These include Winous Point Shooting Club and OttawaShooting Club, private hunt clubs found in thewestern extreme of Sandusky Bay. Other wetlands,such as Pickerel Creek wetland and Willow Pointwetland, are undiked marshes found on the southshore of Sandusky Bay, while Plum Brook wetlandand Sheldon Marsh are undiked wetlands found inthe embayment behind the Cedar Point Amusement Park peninsula.
Hydroperiod
Robb (1989) and Robb and Mitsch (1989, 1992)compared the hydroperiods of a number of dikedand undiked wetlands in the study region duringthe summer of 1988. Relative water levels of thesewetlands (Fig. 9) illustrate a number of points.First, the wetlands that are open to Lake Erie generally dropped in water level through the growingseason due to decreasing lake levels during this period. During this period, the precipitation was
562 W. J. MITSCH
oo
o
8
OctSept
@ 8
AugJuly
Winous Point West
Wmous Point North
@ 80 e0 0 0
e0
May June
50
em 30
10
60
40
em 2
0
-20
Water Quality
Robb (1989) and Robb and Mitsch (1989, 1992)discussed a synoptic sampling and analysis of water quality in a number of wetlands in our studyarea of Lake Erie. Results that compare selectedwater quality parameters in diked and undikedwetlands are shown in Table 2. While no significant differences were inferred for all wetlands inLake Erie, there were significant differences between wetland types for some parameters on a local scale. Turbidity and total phosphorus werehigher in natural (undiked) marshes while conductivity, alkalinity, and orthophosphate were higherin diked or managed marshes. Biologically-available nitrogen (nitrate + nitrite + ammonia) wasabout the same in the two wetland types. The waterquality data suggest that open wetlands, because ofgreater influence by runoff and streamflow, have
OctSept
Sheldon Marsh
Willow Point
AugJulyJune
0 0
00 0 0 em
0 0
0 00 0 00 0
Old Woman Creekem
0
40 Ottawa Allen PondPickerel Creek 0
em 20
0
Plumb Brookem
May
140 +-__-'--_--'----"'-----"'-----"------+
about 750/0 of normal (about 20 cm below normal;Robb 1989). One exception to this trend for managed wetlands was Old Woman Creek wetland,which generally maintained its water level duringthis drought year (see above) due to the formationof a barrier beach at its outflow. Second, the wetlands that are diked generally maintained or evenincreased in water levels due to pumping. Dikedwetlands at Bay View, Ottawa, and Winous Pointreceived pumped water at some times during thesummer. According to Robb (1989), 65 to 75 cmwere added to the Ottawa marshes while 45 to 50cm were added to the Winous Point marshes.
FIG. 9. Patterns of water levels in selected wetlands inlandscape study scale of Ohio's coastal wetlands for: a)natural wetlands (not impounded); and b) diked wetlands. Depths are relative scales and do not representabsolute average depths; multiple data points per dataindicate measurements at several locations. Curves arebest-fit polynomials (from Robb and Mitsch 1989).
50
em 30
10
50
em 30
10
70 0@
em 50
e 030
50
em 30
10
em
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 563
TABLE 2. Comparison of least-square means ofselected water quality parameters in diked and undikedwetlands of Ohio's coastal Lake Erie (Robb 1989).
Diked UndikedParameter Marshes Marshes
temperature, °C 20 22pH 7.67 8.10alkalinity, mg CaCO/L 161* 131conductivity, /Lmhos/cm 1,053* 766turbidity, NTU 26 57*nitrate, /Lg-N/L 330 360nitrite, /Lg-N/L 3 6ammonia, /Lg-N/L 50 40ortho-phosphate, /Lg P IL 73* 55total phosphorus, /Lg P/L 149 237*
*indicates significantly higher number in comparisonbetween diked and undiked marshes, p < 0.01
more suspended sediments and hence more totalphosphorus in the water column than do dikedwetlands which are isolated from such inflows.Lower water levels as a result of the 1988 droughtmade water sampling in natural wetlands difficult,and resuspension of sediments is likely in somecases. Water was generally more plentiful in thediked wetlands because of continual pumping.Furthermore, the presence of organisms that causebioturbation, particularly carp, is more likely inthe open wetlands, as they are specifically controlled in many of the diked wetlands in our studygroup. The higher alkalinity, conductivity, and orthophosphate in diked wetlands suggest that thediked wetlands are like large evaporation basins inthe summer, particularly during a drought, and sa-
linity, particularly as calcium bicarbonate, increases as water with dilute salts is pumped into thewetlands and pure water is lost through evapotranspiration.
Wetland Sediments
Very little has been published on the chemical quality of sediments and their rates of accrual or depletion in coastal wetlands of the Great Lakes. Thesediments of Lake Erie coastal wetlands may bedifferent in chemical and physical characteristics,depending on whether they are more influenced byparent soils, by geologic exchange with Lake Erie,by import from upstream watersheds, or by vegetative productivity in the wetlands themselves. Along history of diking wetlands, thereby isolatingthem from exchange with flooding rivers and theGreat Lakes, may ultimately lead to depauperatesediments and changes in wetland productivity.Several physical and chemical characteristics ofsediments were examined in nine coastal wetlands,five of which are diked wetlands and four of whichare undiked and therefore open to Lake Erie andwatershed exchanges. Table 3 summarizes averageconcentrations of phosphorus, potassium, calcium, magnesium, iron, and percent organic matter. Undiked wetlands appear to have higher concentrations of phosphorus and potassium, whilediked wetlands are higher in calcium and magnesium. With the exception of Sheldon Marsh, whichhad high organic content (to a depth of 33 cm) inits sediments, the undiked wetlands appeared tohave lower organic content (10-12 percent) thatdid the diked wetlands (14-21 percent). This probably reflects the openness of the undiked wetlands
TABLE 3. Selected average sediment concentrations from diked and undiked wetlands of Ohio's coastal Lake Erie.Wetland abbreviations refer to locations on Figure 4. Complete data Mitsch et al. (l989b).
SoilParameter
phosphorus, ppmpotassium, ppmcalcium, ppmmagnesium, ppmiron, ppmorganic content, 010bulk density, glcm3
pHcation exch cap, meq
OSCB
386235,5533288217.50.646.632
Diked WetlandsOSCA WPN BVC BVB
9 43 1 1966 844 99 414,669 4,289 9,783 9,229196 716 137 72119 90 4719.0 14.5 16.5 21.60.91 0.62 0.66 0.557.8 6.1 10.6 10.328 30 50 47
Undiked WetlandsOWC SHM PCK WLP
53 39 28 22868 921 727 8712,480 3,908 6,212 3,633183 278 266 23680 13 109 8710.2 25.5 11.1 11.70.85 0.50 0.97 1.008.1 7.6 8.3 7.115 23 35 22
Average ± std errorDiked Undiked
18±9 35±7515±190 847±426,705± 1,165 4,058±782290± 115 241 ±2185± 13 88±818±1 15±40.67±0.06 0.83±0.118.3±0.9 7.8±0.337±5 24±4
564 W. J. MITSCH
to flooding by inorganic sediments from upstreamwatersheds and export or organic detritus that otherwise would be building organic soils. Concentrations of phosphorus were highest in Old WomanCreek wetland, a wetland that has a significantinput of high phosphorus sediments (see above).The diked Bay View Marshes showed extremelylow concentrations of phosphorus, but no significant differences are noted between the undikedand diked wetlands. These diked wetlands also hadthe lowest concentrations of potassium and magnesium but the highest concentrations of calciumand the highest cation exchange capacity. The pHof the ashed sample solutions was also highest inthe Bay View Marsh samples. All of this suggeststhat the Bay View Marshes are heavily influencedby the limestone geology prevalent in this area ofLake Erie and are dominated by calcium carbonateand bicarbonate.
Wetland Vegetation
Lowden (1969), Meeks (1969), Stuckey (1975,1989), Marshall and Stuckey (1974), Farney andBookhout (1982), Balogh and Bookhout (1989),and Robb (1989) have all documented the vegetation of various coastal wetlands of Lake Erie. Onelandscape-scale question is whether the altered hydrologic conditions and hydrologic isolation imposed by diking wetlands causes an enhancementor loss of vegetative diversity and productivity. Forexample, Stuckey (1975, 1989) has argued that thediversity of native wetland plants along the coastof Lake Erie has been diminished by extensive wetland management such as diking and water levelmanipulation. Results from the study by Robb(1989) appear to indicate that the opposite may bethe case. Although the data are based on relativelyfew sample quadrats in each of the study wetlands,there were 32 aquatic plant species identified in thediked wetland sample stations and only 17 speciesin the undiked wetlands (Robb 1989). But she reported more floating-leaved macrophytes in theopen wetlands because of their ability to adapt tofluctuating water levels. When other measures ofplant net productivity are compared, the dikedwetlands appear to be more productive (Table 4).Net biomass production, as measured by peak biomass in August, is almost twice as high in the dikedwetlands as compared to the undiked wetlands(897 vs. 473 g dry wt m-Z
) and stem counts are overtwice as high (597 vs. 241 stems m-Z
). Biomass isnot an accurate measure of productivity, however,
and the open wetlands may be exporting more oftheir productivity to the lake during the growingseason.
Wetland Remote Sensing and Mapping
Previous studies of remote sensing have evaluatedthe capability to measure wetland variables, andmany efforts have demonstrated remote sensordata as input to wetland studies and wetland inventories (e.g., Shima et at. 1976, Lyon 1981, Mitschet at. 1983, Lyon et at. 1986, Hardisky et at. 1986).Most researchers agree that remote sensing provides a time and cost effective method of mappingwetlands and it can provide various informationsuch as 1) the presence of wetlands, 2) knowledgeof adjacent land use which affect wetland areas,and 3) the change of wetland vegetation over timeassociated with other ecosystem changes. Variousmappings of Great Lakes wetlands have been executed, some by the use of remote sensing and others before remote sensing was generally available.For example, Lyon and Drobney (1984) describedthe area of wetlands as a function of water levels ofLake Michigan from seven sets of black and whiteaerial photographs taken between 1938 and 1977.The data showed that an increase in water level of30 cm resulted in a decrease of 18070 of the 438 haof vegetated wetlands and beaches in the Straightsof Mackinac. These data were subsequently usedto develop a predictive model of long-term effectsof water levels on coastal wetlands (Lyon et at.1986). Balogh and Bookhout (1989) used truecolor 35-mm photographs (slides) taken at 1,500 min 1984 to map the distribution of purple loosestrife (Lythrum salicaria) in Ohio's southwest LakeErie marshes. Visual ground truth by low-altitudeflights was done during the flowering season ofAugust 1985 to improve the accuracy of the maps.Seventy percent of the loosestrife strands werefound within areas designated as wetlands onUSGS topographic maps.
In our study, vertical aerial photographs (slides)were taken of Lake Erie Wetlands on 22 August1988 by the Division of Soil and Water Conservation of the Ohio Department of Natural Resources(Roush et at. 1989). Both 35-mm color and colorinfrared films were used. Approximately 270 photos were taken over several flight lines at an approximate scale of 1:39,000. Interpretation ofthese photographs is not complete, but Figure 10illustrates a preliminary interpretation of a section
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 565
TABLE 4. Selected average macrophyte measurements at peak biomass from diked and undiked wetlands of Ohio'scoastal Lake Erie. Wetland abbreviations refer to locations on Figure 4. Data from Robb (1989).
Average±std error
Measure ofVegetation
Biomass, g dry wt/m2
# species/plot*# stems/m2
OSCB
1,986
1.0170
Diked Wetlands
OSCA WPN WPW
527 590 560
2.0 1.5 3.01,476 92 463
BVC BVB
1,479 239
1.0 1.7474 907
Undiked WetlandsOWC SHM PCK WLP
187 327 859 599
1.0 1.0 1.3 2.3114 177 300 373
Diked(n = 6)
897 ± 277
1.7±0.3597±211
Undiked(n = 4)
473± 149
1.4±0.3
241 ± 59
*only species> 10070 by weight per plot. Plots were 0.5 m2 randomly placed in each wetland (3 to 6 per wetland)
FIG. 10. Example ofa) aerial photograph of landscape-level study ofLake Erie coastal wetlandsand b) subsequent interpreted wetland map (Roush et at. 1989). Photo was taken 22 August 1988by Ohio Department of Natural Resources.
of wetlands in the study area. We chose to usebroad categories such as open water (W), emergentwetlands (E), and floating-leaved wetlands (F). Afew areas are easily recognizable, such as areas of
open water which appear blue or black on the infrared slides. Wooded areas are also relatively easyto recognize by the shapes of the trees. Cattails(Typha spp.) appear dark green on the conven-
566 w. J. MITSCH
Lake Erie WetlandsVickery Quad - OhioFlight 7 - Slides 21-24from Color and IR Aerial Photography byOhio Department of Natural ResourcesAugust 22, 1988
meters
a 200 400 600 800 1000
FIG. 10. b
Idlorneter
@- 0 2 4 6 8
o 1 2 345mile
WETLANDSW open waterE emergent wetlandsE-ct cattailsE-br bulrushE-m millet
F floating-leaved wetlandF-n water lotus
OTHER LAND USES
T forests (trees)X exposed landA agricultural landA-bw buckwheat
D dikes
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 567
Wetland Map of SandU3k.y Ba,
N
Ci J)OJ 4XO Netres fr\1-1 _---'__~I__~_....JI U
T..,r'let Jne;:r.dO'l
ooo
Bare soil
!IIIc:J
FIG. 11. Proposed spatial simulation model ofLake Erie coastal wetlands showing incorporation ofecosystem-levelmodel into geographic mapping system. Map by Gi-Chul Yi.
tional color slides and bright red on the infraredslides, and water lotus (Nelumbo lutea) appearsblue-green on the color slides and light pink on theinfrared slides, but there may be other types ofvegetation which exhibit the same characteristics.Wherever the general category of emergent wetlands (E) was used, the color infrared slides appeared white. R. Kroll, Winous Point ShootingClub, (pers. comm.) suggested that this may indicate the flowering of Sagittaria spp. These mapdata will provide the spatial information necessaryfor building spatial databases and can be used tocalibrate and verify spatial models.
Future Studies - Spatial Modeling
The obvious next step in completing a systems viewof the wetlands in the Sandusky Bay region is tocombine the benefits of dynamic modeling as demonstrated by the ecosystem simulation model ofMitsch and Reeder (1991) with the spatial mappingcapabilities made available from remotely senseddata and geographic information system databases. We hope to develop a dynamic spatial modelof large-scale processes of wetland development.This idea is illustrated conceptually in Figure 11.Such a modeling approach would combine the ben-
568 w. J. MITSCH
efits of recent computer advances in displaying andmanipulating data sets over space with the predictive capabilities of dynamic simulations. Spatialmodeling of coastal wetlands has previously beenapplied to the coastal region of Louisiana (Costanza et af. 1988, 1990) and to the examination ofthe effects of sea level rise on oceanic coastal wetlands (Park et af. 1989). No such approach has yetbeen applied to the Great Lakes, although theyoffer an excellent test case for such models for anumber of reasons: 1) there is a relative abundanceof data on important forcing functions such asstreamflow, water quality, and Lake Erie waterlevels; 2) there is a long history of observation ofthe coastal wetlands along Lake Erie's southernshoreline, with some maps going back to the firsthalf of the 19th century; 3) the fluctuating waterlevels of the Great Lakes, the relatively rapid development of the shoreline, and the current intensemanagement of the wetlands have caused the wetlands of the region to be very dynamic in nature,changing on almost a yearly basis as a result ofmany factors.
REFERENCES
Baker, D. B. 1988. Sediment, nutrient and pesticidetransport in selected lower Great Lake tributaries.U.S. EPA Report EPA-905/4-88-001, GLNOP Report No.1, Washington, D.C.
Balogh, G. R., and Bookhout, T. A. 1989. Purpleloosestrife (Lythrum salicaria) in Ohio's Lake Eriemarshes. Ohio J. Sci. 89:62-64.
Buchanan, D. 1982. Transport and deposition of sediment in Old Woman Creek Estuary of Lake Erie.M.S. thesis, The Ohio State Univ., Columbus, OH.
Costanza, R., Sklar, F. H., White, M., and Day, Jr.,J. W. 1988. A dynamic spatial simulation modelingof land loss and marsh succession in coastal Louisiana. In Wetland Modeling, eds. W. J. Mitsch, M.Straskraba, and S.E. Jf6rgensen, pp. 99-114. Amsterdam: Elsevier.
____ , Sklar, F. H., and White, M. L. 1990. Modeling coastal landscape dynamics. BioScience 40:91-107.
Farney, R. A., and Bookhout, T. A. 1982. Vegetationchanges in a Lake Erie marsh (Winous Point, OttawaCounty, Ohio) during high water years. Ohio J. Sci.82:103-107.
Hardisky, M. A., Gross, M. F., and Klemas, V. 1986.Remote sensing of coastal wetlands. BioScience 36:453-460.
Heath, R. T. 1987. Phosphorus dynamics in the OldWoman Creek National Estuarine Sanctuary-a preliminary investigation. National Oceanic and Atmo-
spheric Administration (NOAA) Technical Memorandum, NOS MEMD 11, Washington, D.C.
Henderson-Sellers, B. 1984. Engineering Limnology.London: Pitman Publishing Ltd.
Herdendorf, C. E. 1987. Coastal wetlands along western Lake Erie: a community profile. U.S. Fish andWildlife Service Report 85(7.9), Washington, D.C.
Hopkinson, C., and Day, J. W., Jr. 1977. A model ofthe Barataria Bay salt marsh ecosystem. In Ecosystem Modelling in Theory and Practice, ed. C. A. S.Hall and J. W. Day, Jr., pp. 235-265. New York:John Wiley and Sons.
International Joint Commission (DC) 1980. Pollutionin the Great Lakes basin from land use activities. DCReport to the Governments of the United States andCanada. International Joint Commission, Windsor,Ontario.
Johnson, M. G. et al. 1978. Management informationbase and overview modelling. PLUARG TechnicalReport No. 002 to the International Joint Commission, Windsor, Ontario.
Kamp-Nielsen, L. 1983. Sediment-water exchange models in Application of Ecological Modelling in Environmental Management, Part A, ed. S. E. Jf6rgensen, pp. 387-420. Amsterdam: Elsevier.
Klarer, D. M. 1988. The role ofa freshwater estuary inmitigating stormwater inflow. Old Woman CreekTechnical Report Number 5. Ohio Department ofNatural Resources Division of Natural Areas andPreserves, Columbus, Ohio.
____ , and Millie, D. F. 1989. Amelioration ofstorm-water quality by a freshwater estuary. Arch.Hydrobiol. 116:375-389.
Kremer, J. N., and Nixon, S. W. 1978. A Coastal Marine Ecosystem. New York: Springer-Verlag.
Lowden, R. M. 1969. A vascular flora of WinousPoint, Ottawa and Sandusky Counties, Ohio. OhioJ. Sci. 69:257-284.
Lyon, J. G. 1981. The influence of Lake Michigan water levels on wetland soils and distribution of plantsin the Straits of Mackinac. Ph.D. dissertation, Univ.of Michigan, Ann Arbor, Michigan.
____ , and Drobney, R. D. 1984. Lake level effects as measured from aerial photos. J. SurveyingEng. 110:103-111.
____ , Drobney, R. D., and Olsen, C. E., Jr.1986. Effects of Lake Michigan water levels on wetland soil chemistry and distribution of plants in theStraits of Mackinac. J. Great Lakes Res.12:175-183.
Marshall, J. H., and Stuckey, R. L. 1974. Aquatic vascular plants and their distribution in the Old WomanCreek Estuary, Erie County, Ohio. ODNR Division ofResearch, Publication DNR-RS-4, Columbus, OH.
Meeks, R. L. 1969. The effect of drawdown date onwetland plant succession. J. Wildl. Mgt. 33:817821.
ECOSYSTEM AND LANDSCAPE APPROACHES TO WETLANDS 569
Meyer, J. L., and R. T. Edwards. 1990. Ecosystem metabolism and turnover of organic carbon along ablack river continuum. Ecology 71:668-677.
Mitsch, W. J., ed. 1989. Wetlands of Ohio's coastalLake Erie: A hierarchy of systems. Final Report,Ohio Sea Grant College Program, Columbus, OH.
____ , and Gosselink, J. G. 1986. Wetlands. NewYork: Van Nostrand Reinhold.
____ , and Kaltenborn, K. S. 1980. Effects of copper sulfate application on diel dissolved oxygen andmetabolism in the Fox Chain of Lakes. Trans. Ill.State Acad. Sci. 73:55-64.
____ , and Reeder, B. C. 1991. Modelling nutrientretention of a freshwater coastal wetland: estimatingthe roles of primary productivity, sedimentation, resuspension and hydrology. Ecological Modelling 54:151-187.
____ , and Reeder, B. C. 1992. Nutrient and hydrologic budgets of the Great Lakes coastal freshwater wetland during a drought year. Wetlands Ecology and Management 1:211-223.
____ , Taylor, J. R., Benson, K. B., and Hill,P. L. 1983. Atlas of wetlands in the principal coalsurface mine region of western Kentucky. U.S. Fish& Wildlife Service Report FWS/OBS 82/72, Washington D.C.
____ , Reeder, B. C., and Klarer, D. M. 1989a.The role of wetlands for the control of nutrients witha case study of western Lake Erie. In Ecological Engineering: An Introduction to Ecotechnology, eds.W. J. Mitsch and S. E. Jf1Srgensen, pp. 129-159. NewYork: John Wiley & Sons.
___ , McNelly, G., and Robb, D. M. 1989b.Physical and chemical characteristics of Lake Eriecoastal wetland sediments. In Wetlands of Ohio'sCoastal Lake Erie: A Hierarchy of Systems, ed.W. J. Mitsch, pp. 135-143, Ohio Sea Grant CollegeProgram, Columbus, OH.
Nixon, S. W., and Oviatt, C. A. 1973. Ecology of aNew England salt marsh. Ecol. Monogr.43:463-498.
____ , Oviatt, C. A., Garber, J. and Lee, V. 1976.Diel metabolism and nutrient dynamics in a saltmarsh embayment. Ecology 57:740-750.
Novotny, V. 1986. A review of hydrologic and waterquality models used for simulation of agriculturalpollution. In Agricultural Nonpoint Source Pollution: Model Selection and Application, eds. A. Girogini and F. Zingales, pp. 9-35. Amsterdam:Elsevier.
Odum, H. T., and Hoskin, C. M. 1958. Comparativestudies of the metabolism of marine waters. Publ.Inst. Mar. Sci. Univ. Texas 5:16-46.
Park, R. A., Trehan, M. S., Mausel, P. W., and Howe,R. C. 1989. Coastal wetlands in the twenty-first century: profound alterations due to rising sea level. InWetlands: Concerns and Successes, pp. 71-80. Amer-
ican Water Resources Association, Bethesda, Maryland.
Pomeroy, L. R., and Wiegert, R. G. 1981. The Ecologyof a Salt Marsh. New York: Springer-Verlag.
Prince, H. H., and D'Itri, F. M., eds., 1985. CoastalWetlands. Chelsea, MI, Lewis Publishers.
Reeder, B. C. 1990. Primary productivity, phosphoruscycling, and sedimentation in a Lake Erie coastalwetland. Ph.D. dissertation, The Ohio State University, Columbus.
Richardson, C. J., and Nichols, D. S. 1985. Ecologicaland analysis of wastewater management criteria inwetland ecosystems. In Ecological Considerations inWetlands Treatment ofMunicipal Wastewaters, eds.P. J. Godfrey, E. R. Kaynor, S. Pelczarski, and J.Benforado, pp. 351-391. New York: Van NostrandReinhold.
Robb, D. M. 1989. Diked and undiked freshwatercoastal marshes of western Lake Erie. Master's thesis, The Ohio State University, Columbus.
____ , and Mitsch, W. J. 1989. Hydroperiod andwater chemistry of diked and undiked wetlands inwestern Lake Erie. In Wetlands of Ohio's CoastalLake Erie: A Hierarchy of Systems, ed. W. J.Mitsch, pp. 113-133. Ohio Sea Grant College Program, Columbus, OH.
____ , and Mitsch, W. J. 1992. Selected chemicalparameters of diked and undiked Lake Erie marshes.In Wetlands of the Great Lakes, Proceedings of aConference. eds. J. Kusler and R. Smardon, pp.130-136. Association of State Wetland Managers,Berne, NY.
Roush, M. J., Robb, D. M., Yi, G. C., and Mitsch,W. J. 1989. Remote sensing of Ohio's wetlands ofwestern Lake Erie. In Wetlands of Ohio's CoastalLake Erie: A Hierarchy of Systems, ed. W. J.Mitsch, pp. 145-157. Ohio Sea Grant College Program, Columbus, OH.
Sager, P. E., Richman, S., Harris, H. J., and Fewless,G. 1985. Preliminary observations on the seicheinduced flux of carbon, nitrogen and phosphorus ina Great Lakes coastal marsh. In Coastal Wetlands,eds. H. H. Prince and F. M. D'Itri, pp. 59-68.Chelsea, Michigan: Lewis Publishers.
Schubel, J. R., and Pritchard, D. W. 1990. Great Lakesestuaries - phooey. Estuaries 13: 508-509.
Shima, L. J., Anderson, R. B., and Carter, V. P. 1976.The use of aerial color infrared photography in mapping the vegetation of a freshwater marsh. Chesapeake Sci. 17:74-85.
Stuckey, R. L. 1975. A floristic analysis of the vascularplants of a marsh at Perry's Victory Monument,Lake Erie. Mich. Bot. 14:144-166.
____ . 1989. Western Lake Erie aquatic and wetland vascular plant flora: its origin and change. InLake Erie Estuarine Systems: Issues, Resources, Status, and Management, NOAA Estuary-of-the-Month
570 w. J. MITSCH
Seminar Series Vol 14, ed. K. A. Krieger, pp. 205256. Washington, DC: V.S. Department of Commerce.
Vollenweider, R. A. 1974. Primary Productivity inAquatic Environments, 2nd edition. IBP HandbookNo. 12, Oxford: Blackwell Sci.
Wiegert, R. G., Christian, R. R., and Wetzel, R. L.1981. A model view of the marsh. In The Ecology of
a Salt Marsh, ed. L. R. Pomeroy and R. G. Wiegert,pp. 183-218. New York: Springer-Verlag.
Yaksich, S. M., Melfi, D. A., Baker, D. B., and Kramer, J. W. 1982. Lake Erie nutrient loads,1970-1980. Lake Erie Wastewater ManagementStudy, V.S. Army Corps of Engineers, Buffalo, NY.
Submitted: 1 May 1991Accepted: 29 June 1992