Functional analysis of a two-year-old created in-stream wetland: Hydrology, phosphorus retention, and vegetation survival and growth

WETLANDS, Vol. 15, No. 3, September 1995, pp. 212-225 ,© 1995, The Society of Wetland Scientists

FUNCTIONAL ANALYSIS OF A TWO-YEAR-OLD CREATED IN-STREAM WETLAND: HYDROLOGY, PHOSPHORUS RETENTION, AND

VEGETATION SURVIVAL AND GROWTH

Steven F. Niswander' and William J. Mitsch: School of Natural Resources

The Ohio State University 202I Coffey Road

Columbus, OH 43210-1085, USA

i Current address: Department of Bioresource Engineering Oregon State University

Gilmore HaIt 116 Corvallis, OR 97331, USA

2 Corresponding author

Abstract: Planted vegetation survival and growth, hydrology, and phosphorus dynamics were monitored for the first two years of wetland development at a 6-ha created riparian wetland, in Franklin County, Ohio, USA. Herbaceous vegetation developed clear zonation and appeared to reduce the invasion of Typha sop. Of the 17 species introduced, 8 were among the 18 most commonly encountered species approximately 2 years after the wetland was constructed. Planted saplings generally survived except where they were found in standing water; Liquidambar styraciflua and Betula nigra were less successful because the site is close to their northern range. Simulation models were developed to predict hydrology, phosphorus retention, and tree growth. The site has a rapidly pulsing hydroperiod driven by runofffrom storm events. Annual simulated phosphorus loading was 17.8 gP m -2 yr-' and retention was 2.9 gP m 2 yr ' (16% of inflow). Fifty years of tree growth simulated for each planted tree species indicated that the site will develop into a forested wetland dominated by Quercus palustris and Fraxinus pennsylvanica with basal area comparable to riparian forests but with lower stem density unless seedlings germinate from the planted trees. Lower water levels in the wetland would favor the development of the woody vegetation over herbaceous vegetation and would also favor higher retention of phosphorus.

Key Words: phosphorus, in-stream wetland, mitigation, wetland creation, freshwater marsh, hydrology, forested wetland, tree survival, ecological model, wetland model, Ohio

INTRODUCTION

Wetlands created or restored to replace the functions of destroyed wetlands in the United States (sometimes erroneously called mitigation wetlands) are designed generally to meet an area requirement based on the amount of wetlands destroyed with less attention paid to specific functional goals. Historically, these wetlands were built and often left to develop on their own with little or no monitoring (Erwin 1991, Mitsch and Cronk 1992, Mitsch and Wilson in press). Follow-up monitoring of such wetlands should identify how the wetlands function and should compare them to some functional standards or reference wetlands. Although these created wetlands probably serve many of the typical wetland functions (nutrient retention, storm water de-

212

tention, wildlife habitat, and ground-water recharge), few studies have been conducted to confirm this. In addition, a study of these wetlands should identify how they might be used to meet many government agencies' goals of increasing wildlife habitat, reduction of nonpoint source pollution, storm water detention, and rec- reation.

Most evaluations of wetlands created for mitigating wetland impact have, in fact, been studies on whether the wetlands were built as designed or studies on the structure, usually vegetation, of the wetland. Sifneos et al. (1992) a n d K e n t u l a et al. ( t 9 9 2 ) looked at a population of created wetlands in southeastern and the northwestern United States, respectively, and found poor areal compensation and poor documentation of the functions of the lost wetlands and the newly de-

Niswander & Mitsch, FUNCTIONAL ANALYSIS OF A CREATED IN-STREAM WETLAND 213

veloped wetlands. Observing and sampling vegetation has usually been the easiest way to infer ecosystem function (see e.g., Wentworth et al. 1988, Jarman et al. 1991, Atkinson el al. 1993, Reinartz and Warne 1993), but vegetation itself is not a good indicator of function (Reinartz and Warne 1993, Mitsch and Wil- son in press). Furthermore, while abundant research has been conducted on water quality functions of wetlands specifically built for wastcwatcr treatment and mine drainage (e.g., Odum et al. 1977, Kadlec and Kadlcc 1979, Ewel and Odum 1984, Fennessy and Mitsch 1989, Hammer 1989, Weidcr 1989, Knight 1990, Brix 1994, Kadlec 1994), less research has been specifically conducted on the hydrologic and water quality functions of wetlands crcaled to mitigate wetland loss (Kusler and Kentula 1990). Much of the in- formation learned from these wastewaler and pollution control wetlands applies to created wetlands (Mitsch and Cronk 1992), but additional research is needed. The Des Plaines Research Wetland Demonstration Project (Hey e ta l . 1989, Sanville and Mitsch 1994, Mitsch et al. 1995) and the Olentangy River Wetland Research Park (Mitsch and Wu 1995) are examples of sites that arc currently being used to investigate the water quality function of constructed wetlands at near- ambient levels of water quality and in experimental conditions.Wetlands built throughout the United States as mitigation for wetlands lost elsewhere provide sim- ilar opportunities for evaluating ecosystem function.

The goal of this study was to evaluate function of a newly created riparian wetland in an urban area in central Ohio. The specific objectives of the study were 1) to evaluate hydrology and planted vegetation survival and growth for the first two years of wetland development, 2) to determine if the wetland was acting as a sink, source, or transformer of phosphorus in early stages of wetland development, and 3) to predict the survival and growth of the planted woody vegetation over a much longer time span.

METHODS

Wetland Design and Construction

The created wetland under study here is located approximately 3 km north of the POll Columbus Inter- national Airport in Franklin County, Ohio, LISA (Fig- ures I and 2). The construction of a 4-ha corporate park in 1992 upstream of the study site led to the filling of approximately 3 ha of degraded wetlands found along a ehannelized stream and in cornfields. Little was doc- umented on the function of those impacted wetlands. The study wetland was built to meet tile reqmrements of the Section 404 program of the Clean Water Act in the United States.

Sedge Meadow/Foresled Erecl ~'nergent

I C ~ , e n Water

Row Oirect ion - -- Vegetation Transects 825 Contours (It. abovo IVSLI ] Water Level ;::~eccrder

Water Samafing Stations

Oodel Poad

2N

/ i

k , l lN 823.3 t

"! i)s

I I 0 5 0 ] O0

[ l l e ~ e r s

" x', ( '~#flow

Figure 1. Site map of study wetland in Franklin County, Ohio, USA, showing vegelation transects, water sampling locations, vegetation zones, and water-level recorder.

A 6.1-ha wetland was built approximately 2 km downstream of the impacted wetland site. The wetland is in an urban setting with three houses located within 100 m. The wetland's watershed, estimated to be 260 ha, contains an industrial park, agricultural land, and private homes. The inflow comes from a stream that is diverted into the wetland during normal tlow conditions; during high flow, excess water can be diverted around tile wetland. The outflow from the wettand flows into a continuation of the same channelized stream that then connects with highway and airport drainage ditches. The hydrology of the area has been grcatly altered by highway construction, stream chan- nelization, tiling of farmland, and industrial development.

The created wetland, built in former agricultural land, consists of two basins (north & south) excavated in the fall of 199 ! to form several zones (Table 1). The created wetland and its surrounding upIand were planted in the spring of 1992 with a wide varicty of vegetation representing 17 herbaceous species and 10 woody species. Each basin consists of an open water zone (t-2 m deep) for over-wintering of fish and mosquito con- lrol. This deep water area is devoid of emergent vas- cular plants. The second zone is a semi-permanently flooded zone planted with emergent vegetation, pri-

214 WETLANDS, Volume 15, No. 3, 1995

Table I. Areas of wetland zones in study wetland shown in Figure 1.

Nol~th Basin South Basin Total Wetland Zone (ha) (ha) (ha)

Open water 0.18 0.17 0.35 Erect emergent 0.21 0.20 0.41 Sedge Meadow 1.67 1.19 2.86 Forested Welland 0.63 1.81 2.44 Total 2.69 3.37 6.06

Figure 2. Aerial photograph of study wetland, August 1993, looking from south to north,

marily Scirpus vatidus and A/isma-Pfantago aquatica. The third zone is floeded only to salurated soils and is designed as a sedge meadow. The fourth zone is a riparian zone that was planted with various woody species at approximately 4.6 m centers at a density of 482 plants/ha. An upland area, which is approximately 50% forested, buffers the wetland from the surrounding landscape. Soil types prior to construction were Ben- nington, Condit, and Pewamo silt loams; all three soils are considered adequate for wetland construction, and both Condit and Pewamo soils are listed in the hydric soils of the United States (U.S. Soil Conservation Ser- vice 198"7).

The created wetland was designed to have in-stream wetland hydrology. To accomplish this, a channelized stream was diverted to flow through the two basins and an outflow weir was constructed to regulate wetland water depth. During the study period, the wetland reccivcd nonpoint-source inflows and high loadings of sediments typical of agricultural and urban pollution.

Monitoring Methods

Hydrology. The size of the watershed was determined to be approximately 260 ha from a U.S. Geological

Survey 7.5-minute topographic map (Northeast Co- lumbus Quadrangle). Aerial photos taken in August 1993 and field reconnaissance were used to verify this estimation. Land uses were noted from aerial photos, and a runoff coefficient was estimated_ Surface flow into the wetland occurred through a channelized stream, and outflow was primarily through an outflow structure. The outflow structure consisted ofa 0.9-m lower and 9-m upper rectangular weir. 1he lower weir contains stop logs which allow for a minimum height range of 0.81 m, as measured from the bottom of the control structure. During low-flow conditions, outflow occurred through the lower weir; during high flow conditions, outflow occurred through both the upper and lower weirs. Outflow from the lower weir was estimated from the Bureau of Reclamation's experimen- tally determined discharge from a l -m contracted rectangular weir (Bellport and Burnett 1984). The upper weir discharge was estimated using the Bureau of Re- clamation's equation for a standard rectangular weir. The head over the outflow weir was measured during each field visit, and an estimation of the depth in the inflow channel was recorded. A Stevens Type F water- level recorder was installed in the south basin in 1993, and its elevation was referenced to an established bench mark. The hydrograph determined from the water- level recorder and rainlhll data obtained from a Na- tional Oceanic and Atmospheric Association (NOAA) weather station were used for calculating the inflow and outflow of the wetland. Design plans for the wetland basins were used to determine the elevation of the weirs, wetland volume, topography, area, and depth. The wetland depths reported from the water-level recorder and hydrology submodel model are measured from the bottom of the outflow structure and do not indicate the average or maximum depth of the wetland.

Climatic data were obtaincd from the NOAA weather station, approximately 3 km from the wetland. Data used in the models include rainfall, temperature, snowfall, and daylight hours. Average monthly tempera- tures for Columbus, Ohio (Ruffner and Bait 1987) were used to calculate potential evapotranspiration according to Thornwaite's equation (Chow 1964).


Vegetation. To evaluate plant survival, diversity, and establishment, biannual field surveys were conducted along three transects in each basin (Figure 1) during October 1992, July 1993, and September 1993. Dom- inant vegetation and any non-vegetated areas were noted. Because woody vegetation was planted at 4.6-m centers, only a limited number of woody plants were encountered along the transects. Plant densities were not measured; however, a qualitative estimate was noted during each field visit by recording vegetation as sparse to abundant.

To evaluate the distribution of woody vegetation, an aerial photo taken in August 1993 was used to locate trees. A sub-sample of approximately one-third of these trees was identified in field surveys and mapped. Spe- cies, diameter at breast height, approximate elevation, and health of each tree were recorded. The height for each of the tree species was calculated according to Botkin (1993):

Hi =riD;) = 137 + b2~D~- b3~Di2 (1)

where, Hi = height of tree (cm), D~ = diameter of tree at breast height (cm), b2, = 2(Hm~,(i) 137)/Dm,~0, b3~ = (Hma~(i) - 137)/Dm,x~02, Hmax(i) = maximum height of species i (cm), and D~(~) = maximum diameter of species i (cm). These tree data were used to estimate the total distribution and the short term survival of the trees planted at the site and provide parameters needed in the tree-growth model.

Water sampling and analysis. Water samples (250 ml) were collected during storm events in the winter of i992-t993 and once every two weeks in the spring of 1993. Four sampling points were established: one at the north basin inflow (A), north basin outflow (B), south basin inflow (C), and south basin outflow (D) (Figure 1). After initial sampling, data showed that stations B and C could be consolidated into one station and water samples were only taken at the north inflow (A), south inflow (C), and south outflow (D). Conduc- tivity was measured with a Hach conductivity meter. The nepholometric method of Standard Method 2130 (APHA et al. 1989) was used to measure turbidity using a Hach ratio turbidity meter.

Ortho-phosphorus samples were also filtered at that time using a 0.45-~m membrane filter. The filtrate was frozen and tested for ortho-phosphorus using a LACH- AT auto-analyzer and molybdate-blue method (Stan- dard Method 4500-P ofAPHA et al. 1989). Unfiltered samples were also frozen and tested for total phosphorus by the same method after persulfate digestion.

Modelling

Hydrology-Phosphorus Model The general approach that was followed for construction of the models is

described by Jorgensen (1986). Parameters were taken from published research, measured in the field, or estimated from field data. The one difference from this method was that a validation step was not conducted due to limited available data. Models were constructed using the higher level simulation language STELLA TM

(Structural Thinking, Experimental Learning Labora- tory with Animation) (Richmond et al. 1987). Two models were built for the site: a hydrology-phosphorus model and a tree-growth model. Because of the highly impermeable soils, ground-water inflow and outflow were considered to be insignificant compared to through-flow in the model. The hydrology-phosphorus model was calibrated using field data collected from January I to August 31, 1993. Simulation time for the hydrology-phosphorus model was 365 days, from Sep- tember 1, 1992 to August 31, 1993. Simulations used a fourth-order Runge-Kutta technique and an integration interval of 0.1 day. Inflow from direct precipitation and surface runoffwas calculated using daily precipitation. A runoff coefficient of 0.90 was assumed in the model. Several simplifications were made in the hydrology submodel. Ground-water infiltration was assumed to be minimal in the overall hydrologic budget because heavy equipment was used for excavation of the wetland basins, causing soil compaction; the soils, by themselves, had very slow infiltration rates when thoroughly wetted. Outflow was assumed to be only through the control structure and by evapotranspiration.

Tree growth model. The tree-growth model was developed to predict the growth and survival of the planted woody vegetation at the study wetland for 50 years, using the average annual conditions for rainfall, growing degree days, nutrient loading, and with different control structure heights (i.e., different water depths). The core of the model was based on Botkin's (1993) optimum tree growth equation:

dD/dt = GiD[ 1 - { D H / D m , x (i)Hm~ gi)}] + [274 + 3b2iD - 4b3iD] (2)

where, D = diameter at breast height (cm), Gi, b2, and b3i = growth parameters for tree species, H = height of tree (cm) from Eqn. 1, Dm~i~ = maximum known diameter for tree species (i) (cm), and H ~ ~i~= maximum known height for tree species (i) (cm).

Growing degree days (DEGD) were also expected to effect tree growth and were calculated according to Botkin (1993) for each species. Water-table depth was also assumed to affect tree growth, and a water factor was calculated using Phipps (1979) equation:

W = 1 - 0.055(T - Wi) 2 (3)

216 W E T L A N D S , V o l u m e 15, No. 3, 1995

Table 2. Planted herbaceous vegetation and naturally colonized species found during the three field surveys at the study wetland, Species in each calego~ are ranked from most to teast common, Plant category is based on Reed (1988),

Number of Times Observed

Plant Oct. July Sept. Species Common Name Category 92 93 93 Total

Planted emergent vegetation AIisma P[antago-aquatica L. Eleocharis spp. Scirpus vatidus Vahl. Carex spp. Scirpusfluviatilis (Torn) Gray

Planted wet prairie vegetation Leersia oo,zoides (L.) Swartz. Verbena hastata L. Calamagrostis canadensis (Michx.)

Beauv. Cyperus strigosus L. Glyceria striata (Lam~) Hitchc. Juncus spp. Ascelepias incarnata L. Aster novae-angliae L. Andropogon gerardi Vitman.

Planted cover crops Setaria sp. Lolium persicum Boiss. and Hohen Phleum pratense L.

Naturally colonized species Daucus carota L. Trifolium pratense L. Typha latifolia L. Ludwigia palustris (L.) Ell. Elymus canadensis L. Potentilla norvegica L. Rumex crispus L. Trifolium hybridum L. Eehinochloa sp. Bidens coronata (L,) Britton. Rudbeckia hirta L. Cirsium arvense (L.) Seop. Galium palustre L. Bidensfrondosa L. Plantago major L. Melilotus officinalis (L.) Pallas Bidens connata Muhl. Ambrosia artemisiifolia L. Aster erleoides L. Erigeron annuus (L.) Persoon Po!vgonum lapathifolium L. Solidago sp. Polygonum pensylvanicum L. Rubus occidentalis L. Scirpus cyperinus (L.) Kunth, Convolvulus sepium L. Penthorum sedoides L.

Water plantain OBL 5 6 6 17 Spike rush OBL 5 6 6 17 Soft-stemmed bulrush OBL 6 5 6 17 Sedge OBL 0 6 6 12 Bulrush, fiver OBL 1 4 2 7

Rice cut-grass OBL 4 6 6 16 Blue vervain FACW+ 0 4 6 10

Blue-joint grass FACW + 0 5 5 10 Straw colored fiat sedge FACW 4 0 3 7 Fowl manna grass OBL 6 0 0 6 Rushes FAC-OBL 0 5 1 6 Swamp milkweed OBL 0 0 3 3 New England aster F A C W - 0 0 1 1 Big bluestem grass FAC 0 0 0 0

Foxtail FACU-FACW 1 6 6 13 Rye grass F A C U - 3 0 0 3 Timothy FACU 1 0 0 1

Queen Anne's lace na 5 6 6 17 Red clover F A C U - 6 6 5 17 Broad-leaved cattail OBL 6 6 5 17 Water purslane OBL 4 6 6 16 Canada wild rye grass FACU + 6 3 6 15 Rough cinquefoil FACU 2 5 6 t 3 Curled dock FACU 1 6 5 12 Alsike clover F A C U - 4 3 5 t 2 Millet FACU-FACW 5 0 6 11 Tickseed sunflower OBL 2 2 6 10 Black-eyed susan F A C U - 0 5 4 9 Canada thistle FACU 1 3 5 9 Marsh bedstraw OBL 3 5 0 8 Beggar ticks FACW 3 1 4 8 Common plantain FACU 0 4 4 8 Yellow sweet clover F A C U - 1 5 1 7 Swamp beggar ticks OBL 1 0 6 7 Common ragweed FACU 0 0 6 6 Many flowered aster FACU 1 0 5 6 Daisy fleebane na 0 5 1 6 Nodding smartweed FACW + 1 0 5 6 Goldenrod FACU-OBL 2 0 4 6 Pinkweed FACW 5 0 0 5 Black raspberry na 0 0 5 5 Wool rush FACW+ 4 1 0 5 Hedge bindweed F A C - 0 0 4 4 Ditch stonecrop OBL 0 0 4 4

Niswander & Mitsch, F U N C T I O N A L ANALYSIS OF A C R E A T E D I N - S T R E A M W E T L A N D 217

Table 2. Continued.

Number of Times Observed

Plant Oct. July Sept. Species Common Name Category. 92 93 93 Total

Fragaria virginiana Mill. Medicago tupulina L. Melilotus alba Medicus Apocynum sibiricum Jacq. Helenium atumnale L Dipsacus sylbestris Hudson Chrysanthemum leucanthemum L. tpomoea sp. Oxalis montana Raf. Penstemon digitalis Nutt. Potamogetgn nodostts Poir, Rosa multfflora Murrax Sparganium sp. Verbascum thapsus L. Vernon& altissima Nutt. Ammania coccinea Roth. Cassia fasciculata Michx. Cuscuta gronov~i Willd. Desmodium cuspidatum (Muhl.)

Loudon Hibiscus trionum L. Lycopus americanus M uhl. Oenothera biennis L. Rorippa istandica (Oed.) Borbfis Rubus aUegheniensis Porter Scirpus atrovirens Muhl. Solanum carolinense L. Xanthium chinense Mill.

Total number of times species observed Total number of different species observed

Wild strawberry FACU 3 0 0 3 Black medick na 0 3 0 3 White sweet clover FACU 0 2 1 3 CLasping-leaved dogbane FAC 0 1 I 2 Sneezeweed FACW+ 0 0 2 2 Teasel FAC 0 1 1 2 Oxeye daisy na 0 2 0 2 Morning glory FACU-FACW 0 2 0 2 Common wood sorrel FAC- 0 0 2 2 White beard tongue FAC 0 1 0 1 Long-leaved pondweed OBL 0 2 0 2 Multiflora rose FACU 0 1 I 2 Bur reed OBL 1 0 1 2 Common mullein na 0 1 1 2 Tall ironweed FAC 0 0 2 2 Ammania OBL 1 0 0 1 Partridge pea FACU 0 0 I 1 Dodder na 0 1 0 1

Dillen's tick trefoil na 0 0 1 1 Flower of an hour OBL 1 0 0 1 Water horehound OBL 0 1 0 1 Evening primrose F A C U - 0 0 1 1 Marsh yellow cress OBL 0 1 0 1 Blackberry FACU - 0 1 0 1 Dark green bulrush OBL 1 0 0 1 Horse nettle na 0 1 0 1 Common c|otbur na 0 0 1 1

106 146 187 439 35 44 50 71

where, W = water factor, T = depth of water table (m), and W~ = o p t i m u m water-table depth for species i.

This equat ion was only used if the annual water level was below the soil surface. I f the annual water depth was above the soil surface, tree growth was assumed to be zero for that year. Growth parameters for each planted species were taken f rom Pearlstine et al. (1985) and Botkin (1993) or calculated using species infor- mat ion f rom Sargent (1933) or Petrides (1988). En- v i ronmenta l factors that effect tree growth were then identified, and the most impor tan t parameters were included in the model . Crowding and shading factors were not included in the model because all o f the trees planted were 5 m apart. Mortali ty in this model can occu~ i f annual growth is less than a given m i n i m u m . In order to carry out simulations, the wetland was divided into the three hydrologic zones where trees were found. The zones were temporar i ly flooded (Zone I), seasonally flooded (Zone II), and semi-permanent ly

f looded (Zone III). The zones were de termined f rom aerial photographs, vegetation surveys, and a simulated hydrograph. A representat ive elevation was chosen to represent each of these zones, and s imulat ions were then carried out for each tree species in each zone. Addi t ional s imulat ions were then conducted that evaluated the effect o f raising and lowering the water level in the wetland. Simulat ions used an Euler 's integration technique with a t ime step o f 1 year and a s imulat ion per iod of 50 years. No validat ion o f the model was possible with data f rom the study site, as site data were only 2-years old and the model was predicting condit ions for 50 years.

R E S U L T S

Ecosystem Structure

Herbaceous Vegetation. Herbaceous vegetat ion surveys showed a noticeable change in the species density,


Table 3. Field data for the planted tree species at the study wetland. Numbers for diameter at breast height (DBH) and height are average +_ standard deviation.

Species Common Name Number DBH (cm) Height (cm) Tree species

Acer rubrum L. Acer saccharinum L. Betula nigra L. Crataegus viridis L. Fraxinus pennsylvanica Marsh. Liquidambar stvrac~llua L. Nyssa sylvatica Marsh. Quercus palustris Muenchh.

Shrubby species Potentilta fruticosa L. Viburnum recognitum Femald

Red maple 30 1.84 _+ 0.20 224 + 9 Silver maple 28 2.36 + 0.50 284 ___ 31 River birch 32 0.81 _+ 0.20 165 + 7 Green hawthorn 39 1.70 _+ 0.40 200 + 15 Green ash 44 2.01 + 0.43 224 _+ 19 Sweetgum 37 1.89 + 0.38 210 + 15 Black gum 25 0.97 _+ 0.26 184 + 13 Pin oak 35 2.12 + 0.41 227 + 18

Shrubby cinquefoil Arrowwood

composition, and zonation (Table 2), with a total of 71 herbaceous species observed during the three field surveys. Thirty-five species were found during the Oc- tober 1992 survey and 50 during the September 1993 survey. Twenty-five of these species were found during both visits. Ten species that were observed in October 1992 were not observed in September 1993, but 25 species observed in September 1993 were not observed in October 1992. Two of the species lost from October 1992 to September 1993 were observed in July 1993. Therefore, there was a net gain of 17 species when the October 1992 and September 1993 surveys are compared. Of the 17 herbaceous species planted (Table 2), 8 were among the 18 most commonly encountered species, 3 were lost or reduced to a number small enough not to be observed, one was never observed, and 6 increased in number but not enough to become common. Of the 3 species lost, 2 were cover crops, timothy (Phleum pratense ) and rye grass (Lotium persicum), while the other was fowl manna grass (Glyceria striata).

A noticeable increase in plant density occurred throughout the site from one sampling to the next. The most obvious change was the dramatic increase in Scir- pus validus, which inhabited a narrow band in each basin in 1992 and completely filled in all shallow water areas by 1993. In addition, an increase in Typha ta- tifolia occurred in the deeper water regions through natural colonization. Zones of vegetation were also observed in the field and from an aerial photograph taken in August of 1993 (Figure 2). A distinct band was observed at the upland border, which is an embankment that surrounds both basins. The next zone was tem- porarily inundated and was dominated by lowland grasses and various field species. Most of the planted trees occurred in this region. The next zone was shallow water to saturated soils and was characterized by Carex spp., Leersia oryzoides, Ludwigia palustris, and Eleo-

charis spp. A narrow band of emergent vegetation followed, with S. validus and T. latifolia intermixed. The open water zone should eventually contain submer- gents, but none were observed during the study period.

Woody Vegetation. Of the 658 trees and shrubs located remotely using aerial 'photos taken in August 1993,270 trees were located and measured in a ground survey meant to be a sub-sample of the entire community (Table 3). Trees and shrubs that had died could not be observed remotely. This was significant because almost all shrubby cinquefoils (Potentilla fruticosa) had either died or were very unhealthy. In addition, eleven of the 270 trees sampled were dead. Potentiflafruticosa accounted for i1% of the planted species, and dead trees accounted for another 4%. Therefore, the initial total number of trees and shrubs was approximately 757.

Modelling Wetland Functions

Hydrology. Figure 3 is a conceptual diagram of the hydrology-phosphorus model, while Table 4 summarizes the model state variables and differential equations. The two state variables for the hydrology submodel are the volume of snow in the watershed (S) and the water volume in the wetland (Q). The model is not concerned with evaluating the snow in the watershed, but this state variable provides the capability of predicting runoff from snow melt. The hydrology submodel was initially built with a limited number of water-level measurements taken in January of 1993 during two storm events.

In April of 1993, a water-level recorder was installed in the south basin, and the model was calibrated against these data (Figure 4). The hydrograph of the site shows a pulsing system, which rises and falls dramatically


Ev apotran spiration ~ - . ~ .

Sediments

- - Ertergy Flow

PhospM.~'us Flow

Figure 3. Conceptual diagram of hydrology-phosphorus model. Symbols according to Odum (1983).

during storm events; the model predicted the depth fairly well with a few exceptions. The differences between the simulated and actual results can be accounted for by several assumptions. First, it was assumed that daily rainfall occurred at a steady rate over twenty- four hours. This assumption would decrease the peak runoffduring any given storm. The second assumption was that the watershed had a constant runoffcoefficient (k~ = 0.9), which was determined using the rational method. This assumption resulted in the model over- predicting runoff during summer when the soil is dry and absorbs much of the rain. However, even with these assumptions, the model provides a relatively good fit. Several major differences not related to the previous assumptions occurred from June I to August 31, 1993. During the latter part of May, vandals removed the stop logs from the weir (C = 0 in the model), causing the water levels to fall dramatically. The control structure variable (C) was not lowered to 0 m in the model, however, because downstream of the outflow structure there is a concrete berm, estimated to be approximately 0.3 m high and 9.1 m wide, which serves as the weir for the wetland. Thus, in the simulation during the period of May 28 to June 22, when the outflow weir was damaged, the outflow was predicted using a 0.91 m rectangular weir, instead of a 9.1 m rectangular weir, resulting in a greater amplitude of the pulsing (Figure 4). After the stop logs were replaced vandals blocked the outflow pipe from the wetland and raised the weir height to 0.81 m, causing the depth to increase approximately 30 cm. The control structure variable (C) was raised in the model to simulate this change and did a good job of predicting the depth. One final result of tampering by vandals was to cause a leak in the outflow logs, causing water levels to drop more than expected at the end of August. Thus, as a result of

Table 4. State variables and differential equations for hydrology-phosphorus model of the study wetland.

Snow as water in watershed, S

dS/dt = k,ASn(t) - k~k~S where,

S Sn(t) A k, k5

Water volume as snow in watershed (m 3) Snowfall (mVday) Area of watershed (m 2) Snow to HzO coefficient f(T) melt coefficient = 0 (if below freezing) =

l (if above freezing) k~ Runoff coefficient

Water volume in wetland, Q

dQ/dt = Pt(t)Aw + Qj(t) - k2Q - k3Q - ET(t) w h e r e ,

Q Pt(t) Q,(0 Aw Aws kl kz d k3 C

Phosphorus

where, P

C,(Q3

Co(Q,)

Water volume in wetland (m ~) Direct precipitation (m/day) = f(Pt(t)) = Pt(t)Aw~k~ = inflow (mVday) Area of wetland (m0 Area of watershed (m0 Runoff coefficient = f(d) outflow over lower weir (day ~) = f(C) & f(Q) head over weir (m) = f(d) outflow over upper weir (day ~) = control structure height (m)

in water column, P

dP/dt = C,(Qt)Q, - Co(Q,)Q

phosphorus in water column and sediments (g)

= f(inflow) phosphorus concentration of inflow (g/m 3)

= f(outflow) phosphorus concentration of outflow (g/m3)

over-predicting runoff and not compensating for the leak, the model over-predicts the depth for the month of August. Even with these differences, the model seems to accurately simulate the hydrology of the site under "normal" conditions, and it was assumed to be valid.

The calibrated hydrology simulation was used to calculate the hydrologic budget for the site from Sep- tember 1992 through August 1993 (Figure 5). The wetland had a rapidly pulsing hydroperiod, and a large amount of water and accompanying material flowed through the system. Surface inflow was the dominant factor in the water budget and was more than forty times as great as direct precipitation. Three simulations were carried out with the control structure raised and lowered to determine its effect on water depth. As the control structure was raised in simulations, the average depth increased; in addition, the amplitude of


129

100

WMer Depth a! Out f low

Strocture ((;m)

20 ~

0

l 1 4 - ' & , | i In , "k-~ .+ ,; ,,,'-' , ,, , ~ , +,: :,,,,I,,,,, ,-t!1. :

,41,~ : . ' , ' ~ ' . n , l a l n~Ion|,l',~ i l k l t l l 2 4 ' - - ~ P Jp~

I I I I

T/me (llilet I

Wetter LOvM Recorder . . . . . . . . Simt l l l l t k )n I

I + FIMd O l t l l . . . . . T ~ of Weir

Figure 4. Simulated wetland water level compared to field data for study wetland for the period of January 1, 1993 to August 31, 1993. This simulation indicates calibration for hydrology submodel.

the storm events decreased, as shown by the standard deviation. The decrease in amplitude was a result of the configuration of the outflow structure and the increased surface area of the wetland. The outflow structure consisted of a lower 0.91-m rectangular weir and an upper 9.1-m rectangular weir. As the height of the control structure was raised, the outflow over the second weir increased, causing a decrease in the amplitude.

Phosphorus Retention. A limited amount of data was collected (Table 5) to evaluate the general water quality of the study wetland. Inflow and outflow data were compared using a non-paired Student t test. This test was used because the water data, while collected at the same time, were not actually paired because of the retention time of the wetland. The only parameter that was significantly different from inflow to outflow was conductivity, which was lower in the outflow. The lack of significant differences in the water samples may have been a result of the sampling times. The hydrology of the site is driven by runofffrom storm events. A single storm may cause the wetland water depth to increase as much as 0.5 m in a 24-hour period. This pulsing hydrology makes it difficult to compare inflow and

Precipitation ~196 61T Evapotranspiration

Surface Inflows j •VlAt = 0 Surface Outflowvs 3870 w I 3904

flows = cmlyr

Figure 5. Annual water budget for study wetland for the period of September 1, 1992 to August 31, 1993.

outflow water quality data with any confidence. In addition, it was often difficult to take water samples during storm events because of the high water level. The north inflow station A was always accessible, but it was often necessary to wade into the wetland to grab water samples at the other stations. Even though extreme care was taken when capturing the samples, it is possible that sediments were stirred up by wading. In addition, the lack of flow data made it difficult to calculate the loading rate. While these data may not give much insight into water quality changes in the wetland, they do provide a baseline for expected values in an urban created wetland.

The water quality data did not provide much insight into the functional values of the wetland. However, using the hydrologic and water quality data, the hydrology-phosphorus model was used to investigate total phosphorus retention in the system. The phosphorus submodel predicted inflow and outflow of phosphorus; therefore, the model could be used to show whether the wetland was a source or sink for phosphorus. When total phosphorus concentrations from field measurements were plotted versus simulated flow, strong correlations were found for inflow and outflow, respectively, with approximately 50 data points for each regression:

Table 5. Summarized water quality data (average + standard error (number of samples)) for study wetland. Sampling occurred from Janua~' 1, 1993 to July 30, 1993. See Figure 1 for sampling locations. Inflow Station A and outflow Station D data were compared using non paired Student I test.

Sta t ion A Sta t ion B S ta t ion C Sta t ion D P a r a m e t e r Inflow Midd le Midd le Outf low

To ta l p h o s p h o r u s (ug/1) O r t h o - p h o s p h o r u s (#g/l) Tu rb id i t y ( N T U ) C o n d u c t i v i t y ( u m h o s / c m )

94 + 13 (52) ' 126 _+ 24 (34) 110 +_ 14 (49) 107 ± 12 (50) L 63 + 7.6(34) 2 69 + 15 (7) 41 + 10 (17) 54 ± 8.6(27)'- 37 +_ 6.9 (51) ~ 50 + 9.3 (35) 48 -4- 6.6 (50) 50 +__ 6.5 (52) ~

462 + 34 (30) ' 458 + 60 (13) 411 _+ 24 (34) 376 + 23 (33)*

' Not significantly different (p = 0.556). 2 Not significantly different (p = 0.196). 3 Not significantly different (p = 0.137). • Significantly different (p - 0.034).

Niswander & Mitsch, FUNCTIONAL ANALYSIS OF A CREATED IN.STREAM WETLAND 221

Total Phosphorus

(rag/I) 1.0 0.9 Inflow O.8 0.7 0.6 0.5 0.4 0.3 ._~=~, [ 0.2 0.1

0

0.9 0.8 Outflow

0.7 0.6

0.4 0.3 OJ' 0.1 i ~

0 |

Field Data

STELLA TM

Simulation

. 1 1/I/93 1/30/93 2/28/93 3/29/93 4/~27/93 5/'26~3

Time (Date)

Figure 6. Simulated inflow and outflow total phosphorus concentrations compared to field data. This simulation indicates calibration for the phosphorus model.

y = (1.0388 x 10-~)x + 6.601 x 10 -5

R 2 = 0.952 (4)

y = (9.6258 x 10 6)x + 9.0659 x 10 -3

R 2 = 0.929 (5)

where, y = total phosphorus concentration, mg-P/l and x = streamflow, m3/day.

The phosphorus regressions seemed reasonable compared to other studies such as Mitsch and Reeder (1991), who calculated a relationship between total phosphorus and flow in the lower Great Lakes tribu- taries. The regression equations 4 and 5 were then used in the model and a simulation was run for January 1 to May 31, 1993 to calibrate the model against the field data. Model simulations seemed to accurately predict inflow and outflow phosphorus concentrations (Figure 6). The model does over-predict both inflow and outflow concentrations during the peaks of large storm events due to either the phosphorus equations used or to the hydrology submodel, which may be over-predicting peak flow.

Using the phosphorus simulation, the annual phosphorus budget was calculated for the period of Sep- tember 1, 1992 through August 31, 1993 (Figure 7). The wetland retained 2.9 gP m 2 yr ~ or 16% of the total phosphorus inflow.

Three simulations were carried out with the control

/ Surface Inflows / Surface Outflows

IOBO (17.8) I &P/&t = 175 (2.9) 905 (14.9)

1 flows = kg-P/yr (g-P/m2-yr)

Figure 7. Annual phosphorus budget for study wetland for the period of September 1, 1992 to August 31, 1993. The total area of the wetland is approximately 6 ha.

structure raised and lowered to determine the effects on phosphorus retention (Table 6). As raising of the control structure was simulated, the annual phosphorus retention decreased. The cause of the decrease is a result of the configuration of the outflow structure. As the height of the control structure was raised, the flow over the second weir increased (9.0 m), causing greater outflow. The greater outflow increased the export of phosphorus. Thus, highest phosphorus retention seemed to occur with the lowest control structure height.

Tree Growth. Three simulations of the growth model were carried out for each species in each of the three zones to determine the effect of different water levels, i.e., there were nine simulations (3 zones × 3 water depths) per each species (Table 7). As one example, Figure 8 shows the growth curve of each tree species over 50 years in Zone II with the annual water depth of 0.60 m. The trees in Zone III, semi-permanently flooded, can only survive in simulation 1 when the average water level was 0.38 m (Table 7). All species in Zone III died when the water level was at 0.60 or 0.72 m. If water level is greater than 0.46 m, the trees' roots are inundated and the trees will not grow. This

Table 6. Phosphorus simulations for study wetland. Initial volume of water (Q) was 9653 m ~ and initial phosphorus was 0.483 kg (0.05 mg/l) for all simulations. The annual inflow of phosphorus for all simulations was 17.8 gP m 2 y r - , . The area of the wetland is 6.1 ha.

Out- Retention Coeffi- flow cient (gP (gP

Simu- C m 2 m-2 lation Description (m) yr -l) yr-') %

1 Stop logs lowcrcd 0.303 14.5 3.4 18.8 2 Stop logs at 1993 0.509 14.7 3.1 17.6

height without removal

3 Stop logs raised 0.661 16.1 1.7 9.5

222 W E T L A N D S , V o l u m e 15, No . 3, 1995

Table 7. Simulation results for study wetland tree-growth model. The diameter at breast height at the end of fifty years is shown. The elevation (E) referenced to the control structure is an average for each zone. The average water depth during the growing season (D~) was calculated from the three hydrology simulations.

Diameter of Trees at Breast Height (D) (cm)

Zone I Zone II Zone III

Dw ~ 0.38 m 0.60 m 0.72 m 0.38 m 0.60 m 0.72 m 0.38 m 0.60 m 0.72 m

Acer rubrum 61.8 61,5 61.2 61.6 61.1 0 (2) 2 61.0 0 (4) 0 (1) Acer saccharinum 43.5 43,2 43.0 43.3 42.9 0 (0) 42.0 0 (0) 0 (3) Betula nigra 30.0 29,7 29.4 29.8 29.3 0 (5) 29.3 0 (0) 0 (2) Crataegus viridis 39.7 39,5 39.3 39.5 39.2 0 (1) 39.2 0 (2) 0 (1) Fraxinus

pennsyfvanica 62.5 61.9 61.4 62.1 61.3 0 (2) 61.2 0 (0) 0 (0) Liquidambar

styraciflua 26.0 25,6 25.3 25.8 25.3 0 (1) 25.2 0 (1) 0 (2) Nyssa sylvatica 55.3 54.9 54.6 55.1 54.5 0 (1) 54.5 0 (0) 0 (3) Quercus patustris 82.2 81.1 80.3 81.4 80.1 0 (5) 80.0 0 (1) 0 (2)

Average water depth during growing season (measured from bottom of control structure). 2 Number of years lree survived before dying is indicated in parenthesis.

simulation is supported by the tree survey, in which 11 trees out of 270 sampled were found dead and all of the dead trees were located in Zone III. The simulated trees in Zone II all survived with an average annual water depth of 0.38 and 0.60 m, but died when the water levels were at 0.72 m (Table 7). If the water depth is above 0.69 m, the trees' roots will be inundated and they will not grow. Finally, the trees in Zone I can survive all three average annual water depths

90

80

70

60

50 DBH, cm

4O

30

20

l0

0

Figure 8.

Q ~ e e c t l $ ])illllsfri.~

F r a x i n u s p e n n s y l v a n i c a

- - - ~ A c e r r u b r u m

, , ,r~'yssc~ s y l ~ ' a t i c a

A c e r s a c c h a r i n u m

. . . . . . . . ( 2 r a t a e g u s v i r~d i s

...... B e r t d a n i g r a

. . . . . L h t t d d a m b a r s~,,mciflu,z

0 10 20 30 40 50 Time, yem-~

Simulated tree growth for selected species at water elevation 0.686 m and an average water depth of 0.60 m. Species are listed from largest to smallest in diameter. DBH = diameter at breast height.

(Table 7) because their elevation is high enough to prevent inundation.

Table 8 summarizes the survival in each of the three simulations. Simulations indicated that all tree species will survive if they are not inundated during the growing season. Differences in land elevation were small enough to not alter the growth greatly. The diameter of trees in different zones at the end of 50 years was often less than 1 cm. This indicates that the model is not very sensitive to the water factor as long as the roots are not completely inundated. The largest trees at the end of 50 years were Quercus patustris, Acer rubrurn, and Fraxinus pennsytvanica (Table 7). The smallest species are Liquidambar styraciflua, Betula nigra, and Crataegus viridis. C. viridis is naturally smaller than the other species and is not affected by the hydrology. L. styraciflua and Betula nigra grew much slower than their optimum, which is accounted for by their low DEGD.

Table 8. Stem density and basal area for each zone at the end of simulated fifty years for trees in study wetland. Total and average basal area for the entire site are shown at the bottom.

Stem

Total Basal Area (mVha)

Density Simu- Simu- Simu- Number Area (Stcms/ lation lation lation

Zone of trees (ha) ha) 1 2 3

l 60 2.44 107 24 24 23 II 304 2.86 106 23 23 0 l l I 44 0.41 107 21 0 0

Average basal area (m2/ha) 22 20 9.5 Total basal area for site (m s) 134 122 57


Tree simulation results, site map, tree survey, and aerial photos were all used to calculate the stem density and basal area at the end of fifty years (Table 8). The maximum simulated basal area for the site was 134 m 2, which was achieved when the water level was lowest (0.38 m). When the water level was maintained at the depth of 0.60 m, total basal area decreased 9 percent to 122 m 2 due to the death of the trees in zone Ill. The overall basal area was predicted to average 22 mVha after 50 years in the low water simulation with approximately 107 stems/ha.

tention time. The high hydrologic loading also caused a high phosphorus loading rate, 17.8 gP m -z yr - ' , much higher than the typical 1 to 4 gP m- 2 yr ~ loading rate ofnonpoint-source wetlands (Mitsch 1992). Con- sidering the short retention time and high phosphorus loading, the wetland appeared to be performing rea- sonably. Second, it is expected that as the wetland matures, increased plant density, reduced soil erosion within the study site, and establishment ofsubmergent plants will increase phosphorus retention, primarily through increased sedimentation.

DISCUSSION

Few studies have evaluated the functions of wetlands created for the mitigation of wetland losses elsewhere (Erwin 1991, Kentula et al. 1992, Mitsch and Gosse- link 1993). This study investigated planted vegetation survival, hydrology, and phosphorus retention and used models to further amplify understanding of hydrology, phosphorus retention, and tree growth at a newly created wetland. This evaluation not only determined some of the functional aspects of this created wetland but also established a baseline data set for future studies.

Hydrology and Phosphorus Retention

The study wetland had a rapidly pulsing hydrograph, driven by runoff during storm events. This type of hydrology is typical for wetlands located in urban watersheds, as these watersheds are often tiled, paved, and channelized, resulting in a high rapid runoff. Water quality data were found to be difficult to analyze by themselves as a result of the rapid pulsing flow-through conditions of the wetland. However, when the data were coupled with a simulation model, phosphorus retention in the urban wetland was estimated to be 2.9 gP m -~ yr -t , a number close to other estimates of phosphorus retention in the literature (e.g., Mitsch and Reeder 1991, Mitsch 1992, Craft and Richardson 1993, Mitsch et al. 1995). The wetland studied here was help- ing to reduce nonpoint-source pollution from urban areas, although the model and field data represent only the first two years of this wetland's existence.

The percent retention of phosphorus in the study wetland (16%) is comparable to some natural flow- through wetlands (e.g., Mitsch and Reeder 1991) but is low compared to some wetlands constructed for controlling nonpoint-source pollution, which may retain up to 90% of the total phosphorus (Mitsch ct al. 1995). The low percent retention, typical of natural flow- through wetlands, appears to be the result of two factors. First, the study site wetland had a high average hydrologic loading, 0.106 m/day, causing a short re-

Herbaceous Plant Success

The planted herbaceous vegetation at the site increased in density during the first two years, and an increase in diversity was observed through the increase in naturally colonizing species. The planting of Scirpus validus appeared to initially decrease the ability of Ty- pha spp. to invade. Planting of this created wetland seems to have accelerated the development of the emergent plant community. Lack of submerged aquat- ic plants suggests that a longer period of time is needed for their establishment; the hydrology may not be suit- able for their introduction~ or there may be a limited seed or propagule bank upstream from the wetland.

Planting a Forested Wetland

Herbaceous vegetation can establish itself relatively quickly and can resemble a natural system within a few years; woody vegetation, on the other hand, takes much longer to establish itself and even longer to ma- ture. l f the site develops into a forested wetland, it will take a few decades before it shows the typical structure of a forested wetland (Mitsch and Gosselink 1993). In general, after two years, trees that were located in the shallow water zone were either dead or severely stressed, trees in the wet meadow portion were healthy, and the trees in the extreme upland were the largest and had the densest foliage. This suggests that during the first year after planting, saplings were stressed by inundation and may have low above-ground productivity and survival rates. A lower water level in this wetland during at least part of the growing season would encourage tree growth and reduce physiological stress. However, the establishment of herbaceous wetland vegetation must also be considered; lower water levels favorable to the woody plants would stress planted emergents and wet prairie species and encourage the establishment of undesirable upland weedy species. The death of the trees in the deep water areas indicates that care must be taken when evaluating hydrology and tree placement in created wetlands. Woody vegetation was generally in good health, with the exception of those


individuals that were inundated. However, almost all observed shrubby cinquefoils (Potentitlafruticosa) were dead; no obvious reason could be found for their death or poor health. Their low survival rate suggests that they may be a poor choice for wetland planting and should not be used in wetland creation projects in central Ohio.

Two other woody species of concern are Liquid- ambar styraciflua and Betula nigra because they are at the border of their northern range. A fluctuating water level, the most typical hydroperiod of this type of freshwater wetland (Mitsch and Gosselink 1993, Kusler et al. 1994), should be the norm ifa combination of woody and herbaceous plants is expected.

Overall, it appears that the tree species chosen for the site will survive and grow, eventually creating a forested wetland with stem density and basal area sim- ilar to a natural riparian system. The initial conclusion from the simulations of tree growth was that the water should be at a very low level for maximum tree growth and survival. However, lowering the water level could drastically alter the herbaceous plant composition and would reduce the wetland area. When the control structure is left at 0.5 m, only a few trees in Zone III (semi- permanently flooded) die. The loss of these trees may be insignificant compared to the damage that may occur to emergent wetland species if the water level is lowered. Therefore, to optimize for wetland area, emergent vegetation growth, and tree survival, the water level should be maintained at an intermediate depth (0.5 m). With this intermediate depth it was predicted that, after 50 years of growth, trees would have a total basal area of 122 m 2 (20 mVha) and stem density of 107 stems/ha. Basal area values are within those reported for natural riparian systems (17.7 to 42 m2/ha), but the stem density is lower than those reported for natural riparian wetlands (370 to 990 stems/ha) (Mitsch and Gosselink 1993). Calculations for this study wetland are based on the model simulations that do not include colonization of new tree species or the birth of seedlings. If these were taken into account, it is expected that the stem density would increase greatly and the basal area slightly. Furthermore, the basal area calculations may be slightly high because there is no chance mortality for the trees in the model. Long-term studies will be needed to validate these model predic- tions.

ACKNOWLEDGMENTS

w e appreciate the support of Marshall Eames and Envirodyne Engineers, Chicago, Illinois for introduc- ing us to the wetland site and partially supporting this study. Ross Laboratories allowed site access. Field studies were assisted by Angela Mehall-Niswander.

Partial salaries and partial support were provided by the School of Natural Resources and the Ohio Agri- cultural Research and Development Center of The Ohio State University. This paper is reprint number 95-006 of the Olentangy River Wetland Research Park, The Ohio State University.

LITERATURE CITED

Amefican Public Health Association. 1989. Standard Methods for the Examination of Water and Wastewater. American Public Health Association, American Water Works Association, and the Water Pollution Control Federation. Washington, DC, USA.

Atkinson, R. B., J. E. Perry, E. Smith, and J. Cairns Jr. 1993. Use of created wetlands delineation and weighted averages as a com- ponent of assessment. Wetlands 13:185-193.

Bellport, B. P. and G. E. Burnett (eds.). 1984. Water Measurement Manual, U.S. Department of the Interior, Bureau of Reclamation. Denver, CO, USA.

Botkin, D. B. 1993. Forest Dynamics: An Ecological Model. Ox- ford University Press, New York, NY, USA.

Brix, H. 1994. Constructed wetlands for municipal wastewater treatment in Europe. p. 325-333. In W. J. Mitsch (ed.) Global Wetlands: Old World and New. Elsevier, Amsterdam, The Neth- erlands.

Chow, V. T. (ed.). 1964. Handbook of Applied Hydrology. Mc- Graw-Hill, New York, NY, USA.

Craft, C. B. and C. J. Richardson. 1993. Peat accretion and N, P, and organic C accumulation in nutrient-enriched and unenriched Everglades peatlands. Ecological Applications 3:446-458.

Erwin, K. A. 1991. An evaluation of wetland mitigation in lhe South Florida Water Management District, Vol. l, Final Report to South Florida Water Management District, West Palm Beach, FL, USA.

Ewel, K. C. and H. T. Odum (eds.). 1984. Cypress Swamps. Uni- versity Presses of Florida, Gainesville, FL, USA.

Fennessy, M. S. and W. J. Mitsch. 1989. Treating coal mine drainage with an artificial wetland. Research Journal of the Water Pol- lution Control Federation 61:t691-1701.

Hammer, D. A. (ed.). 1989. Constructed Wetlands for Wastewater Treatment: Municipal, Industrial and Agricultural. Lewis Pub- lishers, Inc., Chelsea, M1, USA.

Hey, D. L., M. A. Cardamone, J. H. Sather, and W. J. Mitsch. 1989. Restoration of riverine wetlands: the Des Plaines River Wetlands Demonstration Project. p. 159-184. In: W. J. Mitsch and S. E. Jorgensen (eds.) Ecological Engineering: An Introduction to Eco- technology. Wiley, New York, NY, USA.

Jarman, N. M., R. A. Doberteen, B. Windmiller, and P. R. Lelito. 1991. Evaluation of created freshwater wetlands in Massachu- setts. Restoration and Management Notes 9:26-29.

Jorgensen, S.E. 1986. Fundamentals of,Ecological Modelling. El- sevier, Amsterdam, The Netherlands.

Kadlec, R .H . 1994. Wetlands for water polishing: free water surface wetlands, p. 335-349. In W. J. Mitsch (ed.) Global Wetlands: Old World and New. Elsevier, Amsterdam, The Netherlands.

Kadlec, R. H. and J. A. Kadlec. 1979. Wetlands and water quality. p. 436~156. in P. E. Greeson, J. R. Clark, and J. E. Clark (eds.) Wetland Functions and Values: The State of Our Understanding. American Water Resources Association, Minneapolis, MN, USA.

Kentula, M. E., R. P. Brooks, S. E. Gwin, C. C. Holland, A. D. Sherman, and J. C. Sifneos. 1992. Wetlands: An Approach to Improving Decision Making in Wetland Restoration and Creation. Island Press. Washington, DC. USA.

Knight, R. L. 1990. Wetland Systems. p. 211-260. In Natural Systems for Wastewater Treatment, Manual of Practice FD-16. Water Pollution Control Federation, Alexandria, VA, USA.

Kusler, J. and M. E. Kentula. 1990. Wetland Creation and Res- toration: The Status of the Science. Island Press, Washington, DC, USA.


Kusler, J., W. J. Mitsch, and J. S. Larson. 1994. Wetlands. Scientific American 270:64-70.

Mitsch, W.J. 1992. Landscape design and the role of created, restored, and natural riparian wetlands in controlling nonpoint source pollution. Ecological Engineering 1:27--47.

Mitsch, W. J., and B. C. Rcedcr. 1991. Modelling nutrient retention of a freshwater coastal wetland: estimating the roles of primary productivity, sedimentation, resuspension, and hydrology. Eco- logical Modelling 54:151-187.

Mitsch, W. J. and J. K. Cronk. 1992. Creation and restoration of wetlands: some design consideration for ecological engineering, p. 217-259. tn R. Lal and B. A. Stewart (eds.) Advances In Soil Science, Vol. 17. Springer-Vedag New York, Inc., New York, NY, USA.

Mitsch, W. J. and J. G. Gosselink. 1993. Wetlands, 2nd ed. Van Nostrand Reinhold, New York, NY, USA.

Mitseh, W.J.andX.Wu. 1995. OlentangyRiverWetlandResearch Park at The Ohio State University: Annual Report 1994. The Ohio State University, Columbus, OH, USA.

Mitsch, W. J. J. K. C~onk, W. Wu, R. W. Nairn, and D. L. Hey. 1995. Phosphorus retention in constructed freshwater riparian marshes. Ecological ADplications. in press.

Mitsch, WJ. and R.F. Wilson. in press. Improving the success of wetland creation and restoration with know-how, time, and self- design. Ecological Applications

Odum, H.T. 1983. Systems Ecology: An Introduction. Wiley, New York, NY, USA.

Odum, H. T., K. C. Ewel, W. J. Mitsch, and J. W. Ordway. 1977. Recycling treated sewage through cypress wetlands in Florida. p. 35--67. In F. M. D'Itri (ed.) Wastewater Renovation and Reuse. Marcel Dekker, New York, NY, USA.

Pearlstine, L., H. McKellar, and W. Kitchens. 1985. Modelling the impacts of a river diversion on bottomland forest communities in the Santee River floodplain, South Carolina. Ecological Mod- elling 29:281-302.

Petrides, G.A. 1988. Eastern Trees. Houghton Mifflin Co., Boston, MA, USA.

Phipps, R.L. 1979. Simulation of wetlands forest vegetation dynamics. Ecological Modelling 29:257-288.

Reed, P. B., Jr. 1988. National List of Plant Species that Occur in Wetlands: Northeast (Region 1). U.S. Fish and Wildlife Service, Washington, DC, USA, Biological Report 88 (26.1).

Reinartz, J. A. and E. L. Wame. 1993. Development of vegetation in small created wetlands in southeast Wisconsin. Wetlands 13: 153-164.

Richmond, B., S. Peterson, and P. Vescuso. 1987. An Academic User's Guide to STELLA'. High Performance Systems, Lyme, NH, USA.

Ruffner, J. A. and F. E. Bair (eds.). 1987. Fhe Weather Almanac, Gale Research Company Book Tower, Detroit, MI, USA.

Sanville, W. and W. J. Mitsch (eds.). 1994. Creating Freshwater Marshes in a Riparian Landscape: Research at the Des Piaines River Wetland Demonstration Project. Ecological Engineering 3:315-521.

Sargent, C.S. 1933. Manual of the Trees of North America. The Riverside Press, Cambridge, MA, USA.

Sifneos, J. C., E. W. Cake, Jr., and M. E. Kentula. 1992. Effects of Section 404 permitting on freshwater wetlands in Louisiana, Alabama, and Mississippi. Wetlands 12:28-36.

U.S. Soil Conservation Service. 1987. Hydric Soils of the United States. In cooperation with the National Technical Committee for Hydric Soils, Washington, DC, USA.

Welder, R.K. 1989. A survey of constructed wetlands for acid coal mine drainage treatment in Eastern United States. Wetlands 9:299- 315.

Wentworth, T. R., G. P. Johnson, and R. L. Kologiski. 1988. Des- ignation of wetlands by weighted averages of vegetation data: A preliminary evaluation. Water Resources Bulletin 24:389-396.

Manuscript received 12 September 1994; revision received 17 March 1995; accepted 7 April 1995.

Documents

Functional analysis of a two-year-old created in-stream wetland: Hydrology, phosphorus retention, and vegetation survival and growth