The Influence of Hydrology and Time on Productivity and Soil Development of Created and Restored Wetlands

THE INFLUENCE OF HYDROLOGY AND TIME ON PRODUCTIVITY AND SOIL DEVELOPMENT OF CREATED AND RESTORED WETLANDS

DISSERTATION

Presented in Partial Fulfillment of the Requirements for

the Degree Doctor of Philosophy in the Graduate

School of The Ohio State University

Christopher J. Anderson, M.S.

* * * * *

The Ohio State University 2005

Dissertation Committee: Approved by William J. Mitsch, Adviser

Warren A. Dick

P. Charles Goebel Adviser School of Natural Resources

ABSTRACT

In created and restored wetlands, hydrology (the depth, duration, and dynamics of

water in wetlands) and time play an important role in regulating most ecological

processes including productivity and soil development. The influence of hydrology on

created and restored wetlands was examined using full-scale ecosystems and replicated

mesocosm systems at the Olentangy River Wetland Research Park (ORWRP). In one

study, twenty 540-liter tubs or ‘mesocosms’ were planted with either one of two wetland plants common to the region: narrow-leaved cattail (Typha angustifolia L.) or soft- stemmed bulrush (Schoenoplectus tabernaemontani C.C. (Gmel) Palla). For each species, half the mesocosms were pumped with river water based on a monthly pulsing regime while the other half was pumped on a steady-flow regime (an even amount of water was provided weekly). Overall, Typha wetlands were significantly more productive than Schoenoplectus wetlands; however no significant differences in productivity or

morphology were observed between pulsed or steady-flow wetlands among species

groups. No significant differences in nutrient concentrations, uptake or uptake efficiency

were detected among species groups either; however hydrology did influence plant tissue

N:P ratios (P<0.01). For all wetland mesocosms, the mean (±1 SE) N:P ratio was 9.2 ±0.6

for steady-flow and 11.7 ±0.5 for pulsed conditions, suggesting that the steady flow

wetlands were more N limited than pulsed wetlands.

ii A second project evaluated the restoration of flood pulses on a 5.2-ha (13-acre) bottomland forest along the Olentangy River at the ORWRP. In June 2000, the bottomland forest hydrology was restored to approximate natural flooding by cutting

three breeches in an artificial levee constructed between the river and the north section of

the forest and a fourth breech along the natural riverbank at the south section of the forest.

Total aboveground net primary productivity (ANPP) was calculated for two sections of

the forest (north and south) using estimated forest litterfall and wood production. No

significant difference in mean ANPP was detected between the north section (807 ±86 g

m-2 yr-1) and the south section (869 ±56 g m-2 yr-1), however productivity at the north

section was substantially higher than a previous ANPP estimate conducted before

restoration. A notable increase in canopy tree basal growth was noted in the south section

and was primarily due to the prevalence of boxelder (Acer negundo L.) which was the dominant species in this section and showed the same shift in basal area increment during

2003 and 2004.

Soil development over time was evaluated in two experimental wetlands (~1 ha each) that were created in non-hydric soils at the ORWRP in 1994. In May 2004, soil samples were collected (10 years and 2 months after the wetlands were flooded) and compared to samples collected in 1993 (after the wetland basins were excavated but prior to flooding) and 1995 (18 months after the wetlands were flooded). Soils in the two wetlands have changed substantially through sedimentation and organic accretion. Since 1995, soil parameters have been influenced most by the deposition of organic matter from colonized macrophyte communities. Mean percent organic matter at the surface increased from 5.3

±0.1% in 1993 and 6.1 ±0.2% in 1995, to 9.5 ±0.2% in 2004. Mean total P increased

iii from 493 ±18 μg g-1 in 1993 and 600 ±23 μg g-1 in 1995, to 724 ±20 μg g-1 in 2004.

Spatial analyses of percent organic matter (a common indicator of hydric soils) at both

wetlands in 1993, 1995 and 2004 showed that soil conditions have become increasingly more variable. High spatial structure (autocorrelation between data points) was detected in 1993 and 2004, with data in 2004 exhibiting a much higher range of variance between data points (C0 + C: 5.24–9.54) and narrower range of autocorrelation (A0: 24-62 m) than

in 1993 (C0 + C: 0.25–0.49 and A0: 152-239 m).

Sediment accumulation was also evaluated in the same two experimental wetlands.

Higher mean sediment accumulation was detected in the deeper open water zones (62 ±6

and 74 ±5 kg m-2) for the two wetlands than in the emergent vegetation zones (38 ±2 and

39 ±3 kg m-2). Directional spatial structure associated with sediment accumulation was

detected in both wetlands and was attributed to the high accumulation in the open water

zones and the gradual decline in accumulation from inflow to outflow. High

accumulations of Ca (2.4 ±0.2 kg m-2 for both wetlands) and inorganic C (730 ±70 and

-2 717 ±49 g m ) in the OW zones of both wetlands suggest that CaCO3 deposition has

remained a critical process where algae productivity has been highest. Annual rates of

sediment and nutrient accumulation for each wetland were lower than those calculated in

previous years and typically fall between ranges seen for newly created wetlands and

natural wetlands.

Dedicated to Jamie and Sam

ACKNOWLEDGMENTS

There are many individuals that I would like to acknowledge who helped me complete this dissertation. I gratefully acknowledge my adviser William Mitsch for his guidance and support of my research, and the opportunity to study at the Olentangy River Wetland Research Park (ORWRP). I am also grateful to Warren Dick and Charles Goebel for serving on my dissertation committee and providing me with their insight and comments regarding my research- from proposal to completion. I also would like to thank Charles Goebel for lending me several pieces of equipment and software necessary for my bottomland research. I thank Robert Vertrees for serving on my candidacy committee and always keeping me mindful of the policy implications to my research. I would like to thank my fellow ‘Wetlanders’ at the ORWRP, particularly: Anne Altor, Don Bachorowski, Natalie Dillon, Debra Gamble, Dan Fink, Michelle Guthrie, Cheri Higgins, Maria Hernandez, Jeremiah Miller, Amanda Nahlik, Rebecca Swab, Cassie Tuttle, Jan Thompson, and Li Zhang. Aside from helping me with my research in various ways, all of you made coming to the research park fun and enjoyable. A special thanks goes to Li Zhang for helping me out with numerous technical problems over the years. I am also indebted to many of the previous researchers who have worked at the ORWRP, many of whom I have never met. Building on their research is what makes the ORWRP a unique and valuable facility, and indeed I truly felt like I was “standing on the shoulder of giants.” In particular I would like to acknowledge Bob Nairn, Lisa Svengsouk, Sarah Harter, Mathew Cochran, and Michael Liptak for their previous research.

vi Funding for my doctoral research and education came from several sources including: an assistantship from the School of Natural Resources, an Ohio Agricultural Research and Development Center (OARDC) Graduate Research Enhancement Grant, OARDC Grant No. 2002-079, U.S. Department of Agriculture Grant No. 2002-35102-13518, the Ohio Department of Transportation and an Arthur M. Schlesinger Graduate Tuition Fellowship. Finally, I would like to thank my wife Jamie for her love, support and willingness to pick up and move to unknown parts of Ohio. It has been an amazing journey and I am looking forward to sharing the rest of it with you. I love you. I also want to thank the newest star in my life, baby-Sam, who has become an unlimited source of inspiration to me.

vii

VITA

November 19, 1970...... Born - Fairfax, Virginia

1993...... B.S., Forestry & Wildlife, Virginia Tech, Blacksburg, Virginia

1993...... Research Assistant, Indiana University, Mountain Lake Biological Station, Virginia

1993-1994 ...... Environmental Technician, Southern Biomes, Cape Coral, Florida

1994-1996 ...... Environmental Specialist II, Biological Research Associates, Inc., Tampa, Florida

1996-1998 ...... Ecologist, Biological Research Associates, Inc., Tampa, Florida

1998-2001 ...... Sr. Ecologist, Biological Research Associates, Inc., Tampa and Sarasota, Florida

2001...... M.S., Botany, University of South Florida, Tampa, Florida

2002-2005 ...... Graduate Teaching and Research Assistant, The Ohio State University- School of Natural Resources, Columbus, Ohio

viii PUBLICATIONS

Peer-reviewed journal articles

Anderson, C.J., W.J. Mitsch, and R.W. Nairn. 2005. Temporal and spatial development of surface soil conditions at two created riverine marshes. Journal of Environmental Quality 34(6): 2072-2081.

Anderson, C.J. and W.J. Mitsch. 2005. The effect of pulsing on macrophyte productivity and nutrient uptake: a mesocosm experiment. American Midland Naturalist 154(2):305-319.

Anderson, C.J. and B.C. Cowell. 2004. Mulching effects on the seasonally flooded zone of west-central Florida, USA wetlands. Wetlands 24(4):811-819.

Publications in The Olentangy River Wetland Research Park Annual Reports

Mitsch, W. J., C. J. Anderson, M. E. Hernandez, A. Altor, and L. Zhang. 2004. Net primary productivity of macrophyte communities after ten growing seasons in experimental planted and unplanted marshes. In: Mitsch, W. J., L. Zhang, and C. L. Tuttle (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2003, pp. 75-78.

Anderson, C. J. and W. Mitsch. 2004. Physical soil development of two created wetlands at the Olentangy River Wetland Research Park. In: Mitsch, W. J., L. Zhang, and C. L. Tuttle (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 79-84.

Mitsch, W. J., C. J. Anderson, M. E. Hernandez, and L. Zhang. 2003. Net primary productivity of macrophyte communities after nine growing seasons in experimental planted and unplanted marshes. In: Mitsch, W. J., L. Zhang, and C. J. Anderson (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 37-40.

Anderson, C. J. and W. J. Mitsch. 2003. Open-water autotrophs: biomass and distribution in deepwater basins of the experimental wetlands. In: Mitsch, W. J., L. Zhang, and C. J. Anderson (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 41-44.

Anderson, C. J., C. I. Kettlewell, and W. J. Mitsch. 2003. Soil development of two wetland creation areas at the Olentangy River Wetland Research Park in Columbus, Ohio. In: Mitsch, W. J., L. Zhang, and C. J. Anderson (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 51-56.

ix Anderson, C. J. and W. J. Mitsch. 2003. Water chemistry of river flooding at the inflow and outflow of the created billabong wetland at the Olentangy River Wetland Research Park: 1997-2002. In: Mitsch, W. J., L. Zhang, and C. J. Anderson (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 91-94.

Kettlewell, C. I., C. J. Anderson, and W. J. Mitsch. 2003. Soil characteristics of a riparian mitigation wetland (billabong) six years after creation. In: Mitsch, W. J., L. Zhang, and C. J. Anderson (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 95-100.

Anderson, C. J. and W. J. Mitsch. 2003. Vegetation establishment in the mitigation billabong at the Olentangy River Wetland Research Park, 2000-2002. In: Mitsch, W. J., L. Zhang, and C. J. Anderson (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 101-106.

Anderson, C. J. and W. J. Mitsch. 2003. Tree survey at the mitigation billabong at the Olentangy River Wetland Research Park in August 2002. In: Mitsch, W. J., L. Zhang, and C. J. Anderson (eds.), Olentangy River Wetland Research Park at The Ohio State University: Annual Report 2002, pp. 107-110.

FIELDS OF STUDY

Major Field: Natural Resources, specializing in wetland ecology and restoration

TABLE OF CONTENTS

Abstract...... ii

Dedication...... v

Acknowledgments...... vi

Vita...... viii

List of Figures ...... xiv

List of Tables ...... xviii

Chapters:

1. INTRODUCTION...... 1

1.1 Research goals and objectives...... 2 1.2 Literature cited...... 6

2. EFFECT OF PULSING ON MACROPHYTE PRODUCTIVITY AND NUTRIENT UPTAKE: A WETLAND MESOCOSM EXPERIMENT...... 10

2.1 Abstract ...... 10 2.2 Introduction ...... 11 2.3 Methods ...... 12 2.3.1 Experimental design ...... 12 2.3.2 Statistical analyses ...... 16 2.4 Results ...... 17 2.4.1 Hydrology and water depths...... 17 2.4.2 Plant morphology and primary productivity...... 18 2.4.3 Plant tissue nutrient concentration...... 19 2.4.4 Nutrient loading and uptake...... 20 2.5 Discussion...... 21 2.5.1 Plant morphology and primary productivity...... 21 2.5.2 Plant tissue nutrient concentrations and uptake...... 22 2.6 Implications for the full-sized wetland pulsing study ...... 24 2.7 Acknowledgements ...... 25 2.8 Literature cited...... 25 xi

3. THE INFLUENCE OF HYDROLOGIC RESTORATION ON THE PRODUCTIVITY OF A BOTTOMLAND FOREST IN CENTRAL OHIO ...... 38

3.1 Abstract ...... 38 3.2 Introduction ...... 39 3.3 Methods and materials...... 42 3.3.1 Study site...... 42 3.3.2 Climate and hydrology...... 43 3.3.3 Aboveground net primary productivity...... 43 3.3.4 Predicting ANPP, litterfall production and wood production...... 46 3.3.5 Tree-ring analysis...... 47 3.3.6 Predicting basal growth ...... 48 3.3.7 Statistical analyses ...... 49 3.4 Results ...... 50 3.4.1 Hydrology and climate...... 50 3.4.2 Bottomland composition...... 51 3.4.3 Aboveground net primary productivity ...... 52 3.4.4 Predicting ANPP, litterfall production and wood production...... 53 3.4.5 Tree-ring analysis for BAI...... 53 3.4.6 Predicting basal growth ...... 55 3.5 Discussion...... 55 3.5.1 Bottomland productivity...... 55 3.5.2 Relationship between bottomland productivity and flooding...... 57 3.5.3 Forest basal growth before and after restoration ...... 59 3.5.4 Basal growth in response to flooding ...... 60 3.6 Conclusions...... 62 3.7 Acknowledgements ...... 63 3.8 Literature Cited...... 63

4. TEMPORAL AND SPATIAL DEVELOPMENT OF SURFACE SOIL CONDITIONS AT TWO CREATED RIVERINE MARSHES...... 82

4.1 Abstract ...... 82 4.2 Introduction ...... 83 4.3 Methods and materials...... 86 4.3.1 Study area...... 86 4.3.2 Soil sampling design ...... 88 4.3.3 Physical and chemical soil analyses ...... 89 4.3.4 Temporal and geostatistical analyses ...... 90 4.4 Results...... 91 4.4.1 Temporal changes to soil properties...... 91 4.4.2 Spatial characteristics and changes of soil organic matter ...... 93 4.5 Discussion...... 95 4.5.1 Temporal changes to soil properties...... 95 4.5.2 Spatial characteristics and changes of soil organic matter ...... 98 xii 4.6 Conclusions...... 101 4.7 Acknowledgements ...... 101 4.8 Literature cited...... 102

5. SEDIMENT, CARBON, AND NUTRIENT ACCUMULATION AT TWO 10- YEAR-OLD CREATED RIVERINE MARSHES...... 116

5.1 Abstract ...... 116 5.2 Introduction ...... 117 5.3 Methods ...... 120 5.3.1 Study area...... 120 5.3.2 Soil sampling and analyses...... 122 5.3.3 Statistical analyses...... 124 5.4 Results ...... 126 5.4.1 Sediment characteristics and accumulation...... 126 5.4.2 Spatial patterns of sediment accumulation...... 127 5.4.3 Nutrient accumulation...... 128 5.4.3.1 Carbon...... 128 5.4.3.2 Nitrogen, phosphorus and calcium...... 129 5.5 Discussion...... 129 5.5.1 Sediment accumulation and spatial patterns ...... 129 5.5.2 Longitudinal patterns...... 131 5.5.3 Sedimentation rates...... 132 5.5.4 Nutrient accumulation...... 133 5.5.4.1 Carbon...... 133 5.5.4.2 Nitrogen, phosphorus and calcium...... 134 5.6 Conclusions...... 137 5.7 Acknowledgements ...... 138 5.8 Literature cited...... 138

References ...... 153

Appendices Appendix A – Wetland mesocosm plant and river data (2002-2003) ...... 167 Appendix B – Bottomland hardwood forest data (2004-2005) ...... 173 Appendix C – Experimental wetlands soil and sediment data (2004)...... 199

xiii

LIST OF FIGURES

Figure Page

1.1 The Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio...... 9

2.1 Experimental wetland mesocosms used in this study. Cross-section view and dimensions of two mesocosms and associated French drain system. Water depth was controlled by the height of the connected stand pipe extending from the mesocosm tub. Bleed-down orifices were installed at the soil elevation of each mesocosm soil elevation...... 30

2.2 Weekly hydrologic loading rate of pulsed and steady flow mesocosms. Total loading input includes pumped river water and rainfall. Pulsed and steady-flow mesocosms received 698 and 694 cm of river water, respectively, during the 2003 experimental period (April-mid July). Arrows indicate the weeks in which pulsing occurred...... 31

2.3 Mean (± 1 SE) total aboveground and belowground biomass (g m-2) for Schoenoplectus (1-yr), Schoenoplectus (2-yr) and Typha mesocosms at August 2003. Aboveground standard error represents total (live and dead) biomass...... 32

2.4 Mean (± 1 SE) ramet height (cm) and density of Typha and mean (± 1 se) stem length and density of Schoenoplectus (1-yr) and (2-yr) and for pulsed and steady-flow wetlands in September 2002, June 2003, July 2003 and August 2003...... 33

2.5 Mean N and P concentrations (±1 SE) of aboveground plant tissue of wetland mesocosms in August 2003. N:P ratios of <14:1 and >16:1 are indications of N and P limitations, respectively (Koerselman and Meuleman 1996)...... 34

2.6 Mean N retention (±1 SE) for a) Typha and b) Schoenoplectus (2-yr) wetland mesocosms in August 2003. Inflow of NO3-N is based on river water concentrations and pumping rates recorded during the 2003 experimental period (April – August 2003). No significant differences were detected among hydrological regimes for either of the vegetation groups...... 35 xiv

2.7 Mean P retention (±1 SE) for a) Typha and b) Schoenoplectus (2-yr) wetland mesocosms in August 2003. Inflow of P is based on river water concentrations and pumping rates recorded during the 2003 experimental period (April – August 2003). No significant differences were detected among hydrological regimes for either of the species groups...... 36

3.1 Map of the bottomland forest at the Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio, USA indicating site topography and levee breeches (Mitsch and Zhang 2004). Hydrologic restoration was conducted by breaching a levee (Cuts #1-3) along the north section and breaching the river bank at the south section (Cut #4)...... 67

3.2 Experimental layout at the ORWRP bottomland hardwood forest indicating the location and dimensions of tree plots and litter traps. Each tree plot was divided into four quadrants (NW, NE, SW, and SE) for placement of random litter traps including a fifth trap near the plot center...... 68

3.3 Quarterly-annual normal and recorded precipitation totals for Columbus, Ohio based on data collected from the Ohio Agriculture and Development Center weather station (www.oardc.ohio-state.edu/centernet/weather.htm). Precipitation totals reported for January-March, April-June, July-September and October-December of 1991-2004...... 69

3.4 Hydrograph of river water levels (m above MSL) for the Olentangy River for 2001-2004 based on data collected at the Olentangy River Wetland Research Park (Mitsch and Zhang 2004)...... 70

3.5 Aboveground net primary productivity (ANPP), including litter-fall and wood production for a) tree plots in the north, south and upland sections, and b) mean (±1 SE) for north and south section plots for 2004-05. Error bars for the section means represent standard error for ANPP ...... 71

3.6 Linear relationship between the number of days flooded (2003-2004) and aboveground net primary productivity for experimental plots in 2004...... 72

3.7 Linear relationships between topographic variability (log-transformed elevation variance) and a) aboveground net primary productivity and b) wood production for experimental plots in 2004-05...... 73

3.8 Linear relationships between total tree basal area and a) aboveground net primary productivity and b) wood production for experimental plot data in 2004-05...... 74

xv 3.9 Mean (±1 SE) BAI (%) for bottomland canopy trees from the north, south and upland sections from 1991 to 2004. The dashed line represents pre- and post-restoration periods...... 75

3.10 Mean BAI(%) for a) boxelder (Acer negundo L.) and b) Ohio buckeye (Aesculus glabra Willd.) bottomland canopy trees in the flooded and upland sections from 1991 to 2004. The dashed line represents pre- and post-restoration periods...... 76

3.11 Polynomial relationship between a) the number of days of river discharge >154 m3 sec-1 over the preceding two years and basal area increment (BAI) and b) the number of days of river discharge >154 m3 sec-1 and BAI over 2- yr periods from 1991-2004. Open symbols represent post-restoration years...... 77

4.1 The two experimental wetlands at The Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio pumping system and water control structures...... 107

4.2 The 10-m grid and locations used for soil sampling at the ORWRP experimental wetlands in 1993, 1995 and 2004. Shaded areas within the grid map represent approximate location of the deeper, open water (OW) zones...... 108

4.3 Comparison of combined mean (±1 SE) for a) percent organic matter (n=32), b) total C (n=18), c) total P (n=18), d) available P (n=28), e) exchangeable Ca (n=28), f) exchangeable K (n=28), g) exchangeable Mg (n=28), and h) soil pH (n=28) at the ORWRP experimental wetlands in 1993, 1995 and 2004 at 0-8 and 8-16 cm depths. Letters denote differences between years and depths detected at p<0.05 based on Wilcoxon Signed Ranks Test (Bonferroni adjusted)...... 109

4.4 Frequency distribution curves of soil organic matter in 1993, 1995 and 2004 for a) Wetland 1 and b) Wetland 2 based on spatially interpolated data...... 111

4.5 Spatial distribution maps of soil organic matter for Wetland 1 and 2 in a) 1993, b) 1995 and c) 2005. Maps for 1993 and 2004 were generated by ordinary point kriging using isotropic variogram models. Maps for 1995 were generated using inverse distance weighing method for spatial interpolation (see text)...... 112

5.1 Pumping system and water control structures for the two experimental wetlands at The Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio...... 144

xvi 5.2 The 10-m grid and locations used for measuring sediment depth and collecting samples at the ORWRP experimental wetlands in May 2004. Shaded areas within the grid map represent approximate location of the deeper, open water (OW) zones...... 145

5.3 a) Mean (±1 SE) sediment accumulation at the emergent vegetation (EM) zones, open water (OW) zones and total wetland and; b) frequency distribution based on spatially interpolated data from kriging analyses for Wetland 1 and 2. Total wetland sediment accumulation is derived based on weighted average. Kriging data for Wetland 1 and 2 were based on anisotropic analyses (0◦ and 135◦, respectively; see text for variogram results and further details)...... 146

5.4 Spatial distribution maps of sediment accumulations for Wetland 1 and 2. Maps were generated by ordinary point kriging using anisotropic variogram models (0◦ and 135◦, respectively). Degree bearings provided for reference...... 147

5.5 Correlogram (Moran’s I over mean distance) for sediment accumulation in a) Wetland 1 and b) Wetland 2 for isotropic and anisotropic (0◦, 45◦ and 135◦) analyses...... 148

5.6 Mean (±1 SE) accumulation of a) total C, b) organic C, and c) inorganic C for Wetland 1 and 2. Total wetland accumulation based on weighted average...... 149

5.7 Mean (±1 SE) accumulation of a) total N, b) total P, and c) total Ca for Wetland 1 and 2. Total wetland accumulation based on weighted average...... 150

xvii

LIST OF TABLES

Table Page

2.1 Species and hydrology prescribed for wetlands during the experimental hydrology period (April, May and June) in 2003. Actual hydrology period was extended approximately 2 weeks during pump repair. The use of ‘Schoenoplectus (1-yr)’ mesocosms was necessary due to muskrat damage during winter 2002...... 37

3.1 Synopsis of tree plot environmental variables used for regression analyses with forest productivity...... 78

3.2 Importance value (= rel. density + rel. dominance + rel. frequency) of all tree species identified in the north and south sections of the ORWRP bottomland forest. Dominant species (Impt.value >35) are in bold...... 79

3.3 Results of paired t-tests for mean (±1 SE) basal area increment (BAI) (% and cm2 yr-1) of canopy trees pre- and post-restoration. NS denotes non- significant p-value...... 80

3.4 Results of paired t-tests for mean (±1 SE) basal area increment (BAI) (% and cm2 yr-1) pre- and post-restoration for boxelder (Acer negundo L.) and Ohio buckeye (Aesculus glabra Willd.). NS denotes non-significant p- value...... 81

4.1 Variogram characteristics for soil percent organic matter concentrations in Wetland 1 and 2 at 0-8 cm depth for 1993, 1995 and 2004...... 115

5.1 Mean (±1 SE) physiochemical conditions of sediment in the emergent vegetation (EM) and open water (OW) zones of the planted (Wetland 1) and naturally colonized (Wetland 2) wetlands at the Olentangy River Wetland Research Park in May 2004...... 151

5.2 Range of mean annual accumulation rates of sediment and nutrients for Wetland 1 and 2 at the Olentangy River Wetland Research Park, 1994- 2004...... 152

A.1 Aboveground biomass, belowground biomass, total biomass and root:shoot (R:S) ratio estimated from the experimental Schoenoplectus (Sch.) and Typha (Typ.) mesocosms in August 2003...... 168 xviii

A.2 Mean (±1 SE) stem/ramet density for experimental mesocosm in 2002- 2003...... 169

A.3 Mean (±1 SE) number of inflorescence for experimental mesocosms in 2002-2003...... 169

A.4 Mean (±1 SE) maximum stem height length based on the measured length of the five longest stems/ramets at each mesocosm plot ...... 170

A.5 Mean (±1 SE) stem height length based on the measured length of 12 randomly selected stems/ramets at each mesocosm plot ...... 170

A.6 Nutrient concentrations of plant tissue (live and senescent) collected from experimental Schoenoplectus (Sch.) and Typha (Typ.) mesocosms. All plant biomass were harvested in August 2003...... 171

A.7 Olentangy River water P and NO3-N concentration and input into wetland mesocosms during the 2003 experimental wet season...... 172

B.1 Species, importance value, relative density, relative dominance and relative frequency of trees (>5cm dbh) observed in plots at the north section of the bottomland forest ...... 174

B.2 Species, importance value, relative density, relative dominance and relative frequency of trees (>5cm dbh) observed in plots at the south section of the bottomland forest ...... 175

B.3 Cumulative mean leaf litter biomass and monthly section mean (±1 SE) collected in the bottomland hardwood forest leaf traps between June 2003 and May 2004...... 176

B.4 Cumulative mean reproductive material biomass and monthly section mean (±1 SE) collected in the bottomland hardwood forest leaf traps between June 2003 and May 2004 ...... 177

B.5 Cumulative mean woody material biomass and section mean (±1 SE) collected in the bottomland hardwood forest leaf traps between June 2003 and May 2004...... 178

B.6 Tree specific gravity (per Alden 1995 and U.S. Forest Products Laboratory 1974), dbh, tree height and estimated wood production for all trees >5cm dbh in the bottomland hardwood forest tree plots ...... 179

xix B.7 Surveyed elevations of plots corners (NW, NE, SE, and SW) and leaf traps (LT) for each bottomland hardwood forest tree plot. Elevation mean and variance based on all measured plot elevations ...... 190

B.8 Mean trap and plot canopy cover for each plot in the bottomland hardwood forest ...... 191

B.9 Annual tree basal growth increments and mean annual growth increment (±1 SE) for the pre-restoration years (1991-2000) and post-restoration years (2001-2004)...... 193

C.1 Physiochemical soil characteristics at 0-8 and 8-16 cm depths in 2004. Percent organic C results in bold-type were lab-analyzed and those in regular-type were based on regression analysis with percent organic matter. Coordinates based on the 10x10 m grid system at the experimental wetlands and cover type consisted of emergent (EM) and open water (OW) zones (see text)...... 201

C.2 Total C, total N, total P, and pH of experimental wetland soils at 0-8 and 8- 16 cm depths in 2004. Coordinates based on the 10x10 m grid system at the experimental wetlands and cover type consisted of emergent (EM) and open water (OW) zones (see text)...... 204

C.3 Available P, exchangeable cations (Ca, Mg and K) of the experimental wetland soils at 0-8 and 8-16 cm depths in 2004. Coordinates based on the 10x10 m grid system at the experimental wetlands and cover type consisted of emergent (EM) and open water (OW) zones (see text) ...... 206

C.4 Percent and mean (±1 SE) textural classes of sediment in the experimental wetlands in 2004. Sample coordinates based on the 10x10 m grid system at the experimental wetlands; cover type consisted of emergent (EM) and open water (OW) zones; and sub-basin refers to OW zone in proximity to wetland inflow/outflow (see text) ...... 208

C.5 Micronutrient concentrations of sediment (0-8 depth) at the experimental wetlands in 2004. Coordinates based on the 10x10 m grid system at the experimental wetlands and cover type consisted of emergent (EM) and open water (OW) zones (see text)...... 209

CHAPTER 1

INTRODUCTION

In the last 35 years, interest in wetland ecology has increased substantially. This has

come after wide-spread recognition that wetlands provide very valuable services to the

landscape and society (e.g., flood attenuation, water quality improvement and habitat for

wildlife). Perhaps the most important piece of legislation in the United States interpreted

to protect wetlands was Section 404 of the Federal Water Pollution Control Act and other

subsequent amendments [a.k.a. the Clean Water Act (CWA)]. Through a dredge-and-fill permitting process, Section 404 mandated that wetlands be protected for their ability to protect the quality of the nation’s waterways. Administered primarily by the U.S. Army

Corps of Engineers and the U.S. Environmental Protection Agency, policy was implemented that wetland mitigation would be required for wetland impacts that were deemed unavoidable. Therefore starting in the late 1970s the United States entered the foray of creating compensatory wetlands.

Originally there was very little science involved with the design of these wetlands and

not surprising, many of the initial created wetlands were unsuccessful (Erwin 1991).

1 However, as the number of designers trained in wetland ecology grew, there was increasingly more created wetlands that appeared to be comparable to natural wetlands.

However, the criteria used by regulatory agencies to assess successful wetland mitigation have been criticized as being too cursory and the timelines for monitoring too short

(Mitsch and Wilson 1996, Zedler and Calloway 1999). Consequently there is still debate as to how much we are truly replacing through the policy of wetland mitigation and how long does it take. These questions were the motivation for much of this research.

Studying at The Olentangy River Wetland Research Park (ORWRP) (Fig. 1.1), I used created and restored wetlands to evaluate the progression of wetland functions and attributes over time and under variable hydrologic conditions.

1.1 Research goals and objectives

The goal of this dissertation research was to evaluate the role of hydrology and time on wetland productivity and soil development. Using the wetland facilities at the

ORWRP, the specific objectives of this dissertation were the following:

1) to evaluate the influence of a pulsing water regime on the productivity and

nutrient use efficiency of wetland macrophytes (Chapter 2);

2) to examine the effect of hydrologic restoration on the productivity of a

bottomland hardwood forest (Chapter 3);

3) to examine the change in soil physiochemical parameters over the ten-year

history of two experimental wetlands and to examine the spatial changes in the

concentration of soil organic matter (Chapter 4); and

2 4) to evaluate the amount and spatial distribution of sediment and nutrient

accumulation in the same two experimental wetlands (Chapter 5).

Establishing hydrology (the depth, duration, and dynamics of water in wetlands) in a created wetland or reintroducing it to a restored wetland generates conditions that are distinct from terrestrial systems. For instance, a flood-pulse hydrology (the periodic flooding from an adjacent river or lake) that was once common to riverine wetlands in this United States Midwest (Baker et al. 2004), has been shown to increase plant productivity (the production of biomass) and nutrient uptake in forested wetlands throughout the world (Mitsch and Ewel 1979, Junk et al. 1989, Tockner et al. 2000).

There is the potential for this to occur in Ohio as well because many of the inflowing river waters are often laden with high nutrient loads that can have a “fertilizer effect”

(Mitsch and Gosselink 2000). However, when inundation is prolonged, flooding can also cause adverse conditions that may impede plant growth and nutrient uptake (Mitsch and Rust 1994, Kozlowski 1997).

This research included two studies that explored the relationship between flood pulsing and wetland/floodplain productivity. The first study used to 20 replicated mesocosm tubs at the ORWRP (Fig. 1.1) to evaluate pulsing effects on herbaceous wetlands. While pulsing has been documented to enhance forested wetland (Mitsch and

Ewel 1979) and aquatic (Hein et al. 1999) productivity, few studies have evaluated the influence of a pulsing regime on herbaceous riverine wetlands. This study was designed to complement an ecosystem-scale, flood pulsing study conducted concurrently at the 1- ha ORWRP experimental wetlands (Fig 1.1).

A second study evaluated the restoration of a pulsing regime to a 5.2-ha bottomland

hardwood forest at the ORWRP (Fig. 1.1). The effect of hydrology on riparian forest

productivity has been the subject to several studies (Mitsch and Ewel 1979, Taylor et

al.1990, Tockner et al. 2000, Robertson et al. 2001) and many investigators have

concluded that periodic flood pulses have an important and positive effect on the productivity of the ecosystem. This is consistent with Odum’s subsidy-stress model

(Odum et al. 1979), in which flooding can be beneficial to the productivity of the system, depending upon the frequency, timing and duration of the flood events. Other studies however have different results. Brown and Peterson (1983) and Burke et al. (1999) found that permanently flooded forest zones rather than periodically flooded zones had higher productivity while Megonigal et al. (1997) found no difference between upland and periodically flooded forest productivity. Given the inconsistencies in bottomland responses to flooding, it has been suggested that more studies that evaluate an existing forest under a changing hydrologic regime are needed to elucidate the influence of hydrology on bottomland forests (Conner 1994, Megonigal et al. 1997). This is the scenario at the ORWRP bottomland where this research evaluated changes in productivity

4 years after a pulsing hydrology was restored.

Changes in soil condition have been the least studied component of created wetlands.

The anaerobic conditions of wetland soils induced by hydrology have several unique biogeochemical processes that contribute to their ability to absorb pollutants and improve water quality (Mitsch and Gosselink 2000). The organic matter stored in wetland soils is an important food source for invertebrates that become the food base for higher

4 organisms. The organic matter also is a valuable and permanent sink for atmospheric carbon that would otherwise contribute to climate change. Despite the critical importance of soils to overall wetland functions, they are rarely considered when mitigation wetlands are evaluated for success. Research studies that have been designed to compare the soils of created wetlands to natural reference wetlands (Bishel-Machung et al. 1996, Shaffer and Ernst 1999, Zedler and Callaway 1999; Nair et al. 2001, Campbell et al. 2002,

Brooks et al. 2005) typically find some progression toward natural wetland soil

conditions but with substantial deficiencies in many key characteristics (e.g., lower soil

organic matter concentrations, coarser texture, and dissimilar nutrient concentrations and

pH). As part of this dissertation, a survey of soil physiochemical parameters was

evaluated at two 1-ha created marshes at the ORWRP and compared with conditions

observed in 1993 and 1995. Using these data, changes were analyzed over space and

time.

Finally, a separate study using the same two 1-ha wetlands was conducted to evaluate

sediment and nutrient accumulation. Studies that have evaluated conditions in open

wetland systems have shown that newly created wetlands (Fennessy et al. 1994,

Braskerud 2001, Harter and Mitsch 2003) typically report much higher sedimentation

rates than those studying older created wetlands (Craft et al. 2003) or natural wetlands

(Johnston 1991, Peterjohn and Correll 1994, Craft and Casey 2000). Other research has

demonstrated that sediment and nutrient accumulation can vary considerably within

wetlands depending upon preferential flow and proximity to inflows (Reddy et al. 1993,

Mitsch et al. 1995). In a conceptual model developed by Craft (1997), sediment and P

accumulation in created estuarine wetlands is suggested to occur rapidly in the first few

5 years but as it accumulates, the rate of retention eventually peaks and declines, eventually

becoming comparable to natural systems.

1.2 Literature cited

Baker, D. B., R. P. Richards, T. F. Loftus, and J. W. Kramer. 2004. A new flashiness index: characteristics and applications to Midwestern rivers and streams. Journal of the American Water Resources Association 40:503-522.

Bishel-Machung, L., R.P. Brooks, S.S. Yates, and K.L. Hoover. 1996. Soil properties of reference wetlands and wetland creation projects in Pennsylvania. Wetlands 16:532- 541.

Braskerud, B. C. 2001. The influence of vegetation on sedimentation and resuspension of soil particles in small constructed wetlands. J. Environ. Qual. 30:1447-1457.

Brooks, R. P., D. H. Wardrop, C. A. Cole, and D. A. Campbell. 2005. Are we purveyors of wetland homogeneity? A model of degradation and restoration to improve wetland mitigation performance. Ecological Engineering 24:331-340.

Brown S. and D.L. Peterson. 1983. Structural characteristics and biomass production of two Illinois bottomland forests. American Midland Naturalist 110:107-117.

Burke, M. K., B. G. Lockaby, and W. H. Conner. 1999. Aboveground production and nutrient circulation along a flooding gradient in a South Carolina Coastal Plain forest. Can. J. For. Res. 29:1402-1418.

Campbell, D.A., C.A. Cole, and R.P. Brooks. 2002. A comparison of created and natural wetlands in Pennsylvania, USA. Wetlands Ecology and Management 10:41-49.

Conner, W. H. 1994. Effect of forest management practices on southern forested wetland productivity. Wetlands 14: 27-40.

Craft, C.B. 1997. Dynamics of nitrogen and phosphorus retention during wetland ecosystem succession. Wetlands Ecology and Management 4:177-187.

Craft, C. B., P. Megonigal, S. Broome, J. Stevenson, R. Freese, J. Cornell, L. Zheng and J. Sacco. 2003. The pace of ecosystem development of constructed Spartina alterniflora marshes. Ecological Applications 13:1417-1432.

Craft, C. B. and W. P. Casey. 2000. Sediment and nutrient accumulation in floodplain and depressional freshwater wetlands of Georgia, USA. Wetlands 20: 323-332.

6 Erwin, K. L. 1991. An evaluation of wetland mitigation in the South Florida Water Management District, Vol. I. Final Report to the South Florida Water Management District, West Palm Beach, FL, USA.

Fennessy, M. S., C. C. Brueske, and W. J. Mitsch. 1994. Sediment deposition patterns in restored freshwater wetlands using sediment traps. Ecological Engineering 3:409- 428.

Harter, S. K. and W. J. Mitsch. 2003. Patterns of short-term sedimentation in a freshwater created marsh. J. Environ. Qual. 32:325-334.

Hein, T., G. Heiler, D. Pennetzdorfer, P. Riedler, M. Schagerl, and F. Schiemer. 1999. The Danube Restoration Project: Functional aspects and planktonic productivity in the floodplain system. Regulated Rivers 15: 259-279.

Johnston, C. A. 1991. Sediment and nutrient retention by freshwater wetlands: effects on surface water quality. Critical Reviews in Environmental Control 21:491-565.

Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The flood pulse concept in river- floodplain systems. In D. P. Dodge, ed. Proceedings of the International Large River Symposium. Special Issue of Journal of Canadian Fisheries and Aquatic Sciences 106:11-127.

Kozlowski, T. T. 1997. Responses of woody plants to flooding and salinity. Tree Physiology Monograph 1:1-17.

Megonigal, J.P., W.H. Conner, S. Kroeger and R.R. Sharitz. 1997. Aboveground production in southeastern floodplain forests: a test of the subsidy-stress hypothesis. Ecology 78:370-384.

Mitsch, W.J. and K.C. Ewel. 1979. Comparative biomass and growth of cypress in Florida wetlands. American Midland Naturalist 101:417-426.

Mitsch, W. J. and W. G. Rust. 1984. Tree growth responses to flooding in a bottomland forest in northeastern Illinois. Forest Science 30: 499-510.

Mitsch, W. J., J. K. Cronk, X. Wu, and R. W. Nairn. 1995. Phosphorus retention in constructed freshwater riparian marshes. Ecological Applications 5: 830-845.

Mitsch, W. J. and R. F. Wilson. 1996. Improving the success of wetland creation and restoration with know-how, time, and self-deign. Ecological Applications 6:77-83.

Mitsch, W.J. and J.G. Gosselink. 2000. Wetlands, Third Edition. John Wiley & Sons, Inc., New York, NY.

7 Nair, V.D., D.A. Graetz, K.R. Reddy, and O.G. Olila. 2001. Soil development in phosphate-mined created wetlands of Florida, USA. Wetlands 21:232-239.

Odum, E.P., J.T. Finn and E.H. Franz. 1979. Perturbation theory and the subsidy-stress gradient. Bioscience 29:344-352.

Peterjohn, W. T. and D. L. Correll. 1984. Nutrient dynamics in an agricultural watershed; observations on the role of a riparian forest. Ecol. 65: 1466-1475.

Reddy, K. R., R. D. DeLaune, W. F. DeBusk and M. S. Koch. 1993. Long-term nutrient accumulation rates in the Everglades. Soil Sci. Soc. Am. J. 57:1147-1155.

Robertson, A.I., P.Y. Bacon, and G. Heagney. 2001. The response of floodplain primary production to flood frequency and timing. Journal of Applied Ecology 38:126-136.

Shaffer, P.W. and T.L. Ernest. 1999. Distribution of soil organic matter in freshwater emergent/open water wetlands in the Portland, Oregon metropolitan area. Wetlands 19:505-516.

Taylor, J.R., M.A. Cardamone and W.J. Mitsch. 1990. Bottomland hardwood forests: their function and values. p. 14-34. In J.G. Gosselink, L.C. Lee and T.A. Muir (eds.) Ecological processes and cumulative impacts illustrated by bottomland hardwood wetland ecosystems. Lewis, Chelsea, MI, USA.

Tockner, K., F. Malard and J.V. Ward. 2000. An extension of the flood pulse concept. Hydrological Process 14:2861-2883.

Zedler, J.B. and J.C. Callaway. 1999. Tracking wetland restoration: do mitigation sites follow desired trajectories? Restoration Ecol. 7:69-73.

Fig. 1.1 The Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio.

CHAPTER 2

EFFECT OF PULSING ON MACROPHYTE PRODUCTIVITY AND NUTRIENT

UPTAKE: A WETLAND MESOCOSM EXPERIMENT

2.1 Abstract

A study was conducted to evaluate the effect of a pulsing hydrology on the

productivity and nutrient uptake of an herbaceous, riverine wetland. Pulsing effects were

evaluated using twenty 0.9-m2 wetland mesocosms: 10 planted with Schoenoplectus

tabernaemontani (C.C. Gmel) Palla and the other 10 planted with Typha angustifolia L.

For each species, half the mesocosms were subjected to a 3-month pulsing regime while

the others were subjected to steady-flow conditions. Hydrology parameters were selected

to approximate a pulsing experiment being carried out concurrently at two 1-ha wetlands

at the research site. Typha wetlands were significantly more productive than

Schoenoplectus wetlands; however no significant differences in productivity or morphology were observed between pulsed or steady-flow wetlands among species groups. No significant differences in nutrient concentrations, uptake or uptake efficiency

were detected among species groups either, however hydrology did influence plant tissue

10 N:P ratios. For all wetland mesocosms, the mean (±1 SE) N:P ratio was 9.2 ±0.6 for steady flow wetlands and 11.7 ±0.5 for pulsed, suggesting that the steady flow wetlands were more N limited than pulsed wetlands. The potential applications and limitations of applying these results to the 1-ha wetlands study are discussed.

2.2 Introduction

Wetland plant productivity and nutrient uptake can be influenced by hydrologic parameters including flood depth (Waters and Shay 1992, Newman et al. 1998, Kellogg et al. 2003), flood duration (Newman et al., 1998) and flooding frequency (Giovannini and

Da Motta Marques 1998, Tanner 1999, Casanova and Brock 2000). Further, water level fluctuations associated with a pulsing hydrology may increase wetland productivity and nutrient uptake (Mitsch and Ewel 1979, Mitsch et al. 1979, Junk et al. 1989, Odum et al.

1995, Day et al. 2000). Inflowing flood pulses often contribute higher than normal concentrations of nutrients thus providing a fertilizing effect for plant growth (Spink et al.

1998). Increased nutrient availability can also occur in wetland sediments by the release of phosphorus during anaerobic conditions and the increase in nutrient mineralization caused by the fluctuation of wet-dry soil conditions. However, prolonged wetland inundation increases the potential for anoxic conditions that can be detrimental to macrophyte vegetation (van der Valk and Davis 1978).

Pulsing water regimes are characteristic of riparian wetlands adjacent to flashy streams or rivers in the Midwest United States (Baker et al. 2004). The temporary nature of river flood pulses often leads to rapidly dropping water levels, thus decreasing the chance for soil anoxia while increasing nutrient mineralization and the potential for

11 greater productivity (Mitsch and Rust 1984). The influence of a pulsing hydrology on

increased productivity has been well demonstrated for wetland forests (Mitsch et al. 1991,

Brown 1981) and planktonic communities (Hein et al. 1999), but has been less

predictable for herbaceous wetlands. Because of shallow rooting, herbaceous wetland vegetation may be more sensitive to different flood conditions (depth, duration, etc.) that

may make a pulsing effect more difficult to detect.

The purpose of this study was to determine if a pulsing hydrologic regime would elicit

morphological and functional responses by two common wetland plants in experimental

mesocosms. This project was conducted in conjunction with a multi-year experiment

using two 1-ha created wetlands to examine the effects of a pulsing regime. We

hypothesized that productivity and nutrient uptake would be higher in the pulsed water

regime, and that pulsing river water into wetlands would result in greater N uptake by

vegetation compared to a steady flow regime.

2.3 Methods

2.3.1 Experimental design

The project was conducted at the 12-ha Olentangy River Wetland Research Park

(ORWRP) at The Ohio State University in Columbus, Ohio, USA (latitude N40.021◦, longitude E83.017◦). A total of twenty mesocosm tubs (Fig. 2.1; area=0.9 m2, volume=

540 liters) were used. In each mesocosm a total of 8-10 cm of pea-size gravel was placed

in the bottom to allow seepage. Approximately 30 cm of on-site alluvial, upland soil

(Ross and Eldean series, consisting of silt loam, silt clays and clay loams; Mcloda and

Parkinson 1980) was placed on top of the gravel. Water levels were controlled with a

12 French drain system established for each mesocosm using 5-cm diameter pvc pipes (Fig.

2.1). The control structure (top of the stand pipe) was established 5 cm above the soil

layer. To provide gradual water subsidence, a bleed down orifice (2 mm hole) was drilled

into the side of each pipe at the mesocosm soil elevation.

To evaluate the effect of pulsing, mesocosms were randomly assigned to one of four treatments based on macrophyte species and hydrology (Table 2.1). In March 2002, a total of 30 rhizomes each of Typha angustifolia L. (Cooperrider et al. 2001) (narrow-

leaved cattail, hereafter Typha) and S. tabernaemontani (C.C. Gmel) Palla (Cooperrider et al. 2001) (soft-stemmed bulrush, a.k.a. Scirpus validus, hereafter Schoenoplectus) from

a Midwestern nursery were trimmed to approximately 25 ±5 g in size and planted 3 cm

below the soil surface. Three rhizomes (with no inter-species mixing) were planted in

each mesocosm. To expedite growth of the planted rhizomes, wetlands were kept in

moist condition throughout the summer and autumn 2002 with groundwater equally

distributed to all mesocosms. At the end of August 2002, initial condition aboveground

biomass was non-destructively estimated in each mesocosm by relating various

morphological parameters to reference plants. For Schoenoplectus, stem density, mean

stem height, number of flowers, and mean maximum stem height were recorded. For

Typha, the number of leaves per ramet, ramet density, mean ramet height, number of

flower spikes, and mean maximum ramet height were recorded. Over the winter of 2002-

2003, the mesocosms were not watered; a layer of snow and ice covered them throughout

most of the season.

In March 2003 it was discovered that muskrats (Ondatra zibethicus) excavated 3

Schoenoplectus wetlands of nearly all its rhizomes during winter 2002-03. In April 2003,

13 sod containing Schoenoplectus rhizomes were collected from control wetlands used in another mesocosm experiment, and transplanted into the disturbed mesocosms. Each disturbed mesocosm received five pieces of Schoenoplectus sod that were approximately

15 x 20 cm wide and 4 cm thick. All measurements from these mesocosms were analyzed separately and designated as ‘Schoenoplectus (1-yr)’ wetlands, compared to

‘Schoenoplectus (2-yr)’ for the undisturbed wetlands (Table 2.1).

During the experimental hydrology period (April through June 2003), water was pumped from the Olentangy River based on the pulsing schedule in Table 2.1. By design, all mesocosms received a near equivalent monthly hydrologic load of 160 cm month-1.

Water was pumped directly from the Olentangy River to the mesocosms via a low-

pressure pump and an elevated reservoir tank system. A garden hose was used for each

water regime and extended along each mesocosm. At each nexus point, the hose was

attached to the mesocosm rim and a volume-adjustable irrigation sprayer (Raindrip®

R180C) was used to control the amount of water pumped into each mesocosm. In mid-

May 2003, the pumping schedule was postponed for 11 days to repair the pumping system.

Pumping volumes and water depths were recorded for each water regime. Three mesocosms were omitted from the study in 2003 because of faulty drainage systems that developed and could not be repaired. During the experimental hydrology period, a 250- mL sample of the river water was collected weekly and analyzed for NO3-N and total P

using a Lachat QuikChem IV automated system and Lachat methods (U.S. EPA 1983).

Total P was analyzed using the ascorbic acid and molybdate color reagent methods after

digesting with 0.5 ml of 5.6N H2SO4 and 0.2 g NH3SO4 to 25ml of sample and exposing

14 the samples to a heated and pressurized environment for 30 minutes in an autoclave.

Using the Lachat automated system, nitrate was analyzed using the cadmium reduction

method. Nutrient concentrations were used with pumping rates to estimate weekly NO3-N and total P nutrient loads into the mesocosms (converted to g m-2 for comparison).

Plant morphology was measured on 18 June, 23 July and 28 August 2003 using the

same parameters measured in September 2002. After the August 2003 measurements, all

aboveground vegetation in each mesocosm was harvested and the mesocoms were

covered until belowground biomass was harvested. Harvested aboveground material was

air-dried and a subsample was oven dried for 3 d (or until constant mass was achieved) at

80◦ C. Total aboveground biomass was calculated and converted to g m-2 for comparison.

In September 2003, soil was extracted from each mesocosm and carefully washed from all root material. Root material was weighed for each mesocosm and a subsample was air dried for 5 d at 105◦ C or until constant mass was achieved. Total belowground

biomass was estimated for each mesocosm and converted to g m-2 for comparison.

Belowground biomass data was combined with aboveground data in August 2003 to

calculate total biomass and the mean root:shoot ratio.

To estimate tissue nutrient concentrations for each mesocosm, 10 mature stems were

randomly selected and analyzed for nutrient content. Tissue specimens were air dried,

ground to pass through a 2 mm sieve, and mixed to make a homogenous sample. In

mesocosms where significant plant senescence occurred (>50% of the plant surface

yellowed), 10 random senescent stems were analyzed separately from the living stems

and because of the potential for nutrient translocation, the belowground tissue was

analyzed separately by randomly selecting five 5-cm root/rhizome sections. All tissue

15 specimens were sent to Service Testing and Research (STAR) Laboratory, Ohio

Agricultural Research and Development Center, Wooster, Ohio and analyzed for total N

and P. Samples were digested with HClO4/HNO3 and analyzed for total P by inductively

coupled plasma emission spectrometry (Isaac and Johnson 1985). Samples were analyzed

for total N through combustion analysis (AOAC 1989). Using mean nutrient

concentrations, the mean N:P ratio of the aboveground tissue was calculated for all treatment types and compared to thresholds developed by Koerselman and Meuleman

(1996) to detect possible community level N or P limitations. As an indication of site fertility, tissue nutrient concentrations were used with peak biomass measurements in

August 2003 to calculate total nutrient uptake for each wetland.

2.3.2 Statistical analyses

For each species group [Typha, Schoenoplectus (2-yr) and Schoenoplectus (1-yr)], a two-way, repeated measure analysis of variance (ANOVA) was used to examine for differences in mean macrophyte morphological measurements between pulsed and steady-flow wetlands. An independent t-test was used to compare mean productivity,

-1 nutrient uptake and nutrient loads (mg total P and NO3-N wk ) between the pulsed and

steady-flow mesocosms. In the case of root:shoot ratios and leaf-tissue N and P

concentrations, Schoenoplectus (2-yr) and (1-yr) data were comparable (P>0.05, t-test)

and therefore pooled. A two-way ANOVA was conducted to compare the factors of

hydrology (pulsed and steady flow wetlands), species, and species x hydrology interaction. Data were analyzed and transformed when necessary to meet assumptions for

parametric statistics. For all comparative tests, P-values <0.05 were considered

16 significant and P-values <0.01 were considered highly significant. Tests for parametric statistic assumptions and t-tests were conducted using Minitab Release 14 (Minitab, Inc.

2003), and two-way ANOVAs were conducted using Systat v.10.2 (Systat Software Inc.

2002).

2.4 Results

2.4.1 Hydrology and water depths

River water was pumped into the wetland mesocosms for 4.5 months in 2003. Three week-long pulses of river water occurred during the experimental hydrology period

(between 3 April and 21 July 2003, Fig. 2.2) to simulate spring and early summer flood pulses typical of the region. From 22 July and 18 August 2003, all mesocosms received an equivalent load (approximately 32 cm wk-1) (Fig. 2.2). Water levels fluctuated based on the prescribed hydrologic loading rates and other environmental factors including rainfall, evapotranspiration, and apparent differences in permeability. In the pulsed wetlands, weekly mean water depths were 6.7 ±2.4 cm during the non-pulsing weeks and rose to 15.3 ±0.8 cm during the three pulsing weeks. The steady flow wetlands maintained a more consistent mean water level of 11.2 ±1.1 cm. Based on the measurements taken during the experimental hydrology period, water levels were at or below the surface (<1 cm) 16.4% of the time for pulsed wetlands and 6.9% of the time for steady-flow wetlands. Mean water depths of all mesocosms progressively decreased throughout the growing season as temperatures increased and as evapotranspiration increased.

2.4.2 Plant morphology and primary productivity

There was no significant difference in peak dry-weight biomass between the pulsed

and steady-flow mesocosms for Typha or Schoenoplectus (1-yr and 2-yr) (Fig. 2.3;

Appendix A, Table A.1). The highest mean aboveground biomass was recorded in the

steady-flow Typha wetlands (1032 ±53 g m-2) while pulsed Schoenoplectus (1-yr)

wetlands had the lowest (296 ±58 g m-2). Pulsed Typha wetlands had the highest

belowground biomass (1551 ±160 g m-2) and highest overall (above and belowground) biomass (2557 ±184 g m-2) while pulsed Schoenoplectus (1-yr) wetlands had the lowest

overall biomass (800 ±69 g m-2).

By early August, all Schoenoplectus wetlands had some senescence of mature culms,

particularly in the 2-yr, steady-flow wetlands (Fig. 2.3). By the end of August, the mean

dry weight of senescent biomass in the 2-yr, steady-flow wetlands was 234 ±89 g m-2, or approximately 38% of the live biomass dry weight. A comparable distribution (32%) occurred in the pulsed Schoenoplectus (2-yr) wetlands. The Schoenoplectus (1-yr) wetlands had lower percentages of senescent biomass for both steady and pulsed wetlands

(14% of the live dry weight for both wetland types). Between July and August new

Schoenoplectus shoots had emerged; however a substantial amount of the aboveground stems had yellowed by the end of August 2003.

The mean root:shoot ratios for all species groups ranged between 1.4 (steady-flow

Typha wetlands) and 1.7 (steady-flow Schoenoplectus [1 yr] wetlands) in August 2003.

No differences in root:shoot ratios were detected when the factors hydrology (F

18 1,13=0.021, P=0.89), species (F 1,13=0.934, P=0.35) or species x hydrology (F 1,13=0.800,

P=0.39) were analyzed.

No significant differences were detected in plant morphology measurements between pulsed and steady-flow wetlands for all three species groups. Mean Typha ramet density was highest for pulsed wetlands in August 2003 with 97 ±4 ramets m-2; however both

pulsed and steady-flow wetlands averaged over 95 ramets m-2 by mid-July (Fig. 2.4;

Appendix A, Table A.2). Pulsed and steady-flow Typha wetlands also had similar

monthly trajectories in mean ramet height (Fig. 2.4), mean number of flower spikes, mean

number of leaves/ramet, and mean maximum ramet heights (Appendix A, Table A.3-5).

Schoenoplectus (1-yr and 2-yr) wetlands were more variable than Typha between pulsed

and steady-flow wetlands but no significant differences were detected (Fig. 2.4).

Senescence of mature stems between July and August reduced mean stem density and

stem height of all Schoenolectus (2-yr) wetlands. Despite the higher percentage of senescence, Schoenoplectus (2-yr) wetlands still had greater mean stem length than the

Schoenoplectus (1-yr) wetlands.

2.4.3 Plant tissue nutrient concentration

Nutrient concentrations were comparable between Schoenoplectus (1-yr) and (2-yr), and therefore were combined to analyze hydrological regime effects on plant tissue nutrient concentrations (Appendix A, Table A.6). Analysis of plant tissue indicated that there were no significant differences in N or P concentrations for pulsed and steady-flow

Typha or Schoenoplectus wetlands. However, when N:P ratios were analyzed, a significant difference was detected for hydrology (pulsed and steady-flow wetlands) (F

19 1,13=9.95, P<0.01). No differences in N:P ratios were detected when species (F

1,13=0.021, P=0.89) or species x hydrology interaction (F 1,13=0.463, P=0.51) were

analyzed. For all mesocosms combined, the mean N:P ratios were 9.2 ±0.6 for steady

flow wetlands (9.1 ±0.7 for Schoenoplectus and 9.5 ±1.0 for Typha) and 11.7 ±0.5 pulsed

wetlands (12.1 ±0.5 for Schoenoplectus and 11.5 ±0.8 for Typha). For senescent

Schoenoplectus tissue, no difference was detected in N and P concentrations between

steady-flow (11.0 ±0.5 mg N g-1 and 0.9 ±0.1 mg P g-1) and pulsed (12.2 ±0.9 mg N g-1 and 1.0 ±0.0 mg P g-1) wetlands. Using the N:P ratios calculated for each hydrology and

species type, a comparison was made to threshold values determined by Koerselman and

Meuleman (1996) (Fig. 2.5). In all cases, the N:P ratios were less than the 14:1 ratio

which indicates that the wetlands were nitrogen limited.

2.4.4 Nutrient loading and uptake

A weekly mean nutrient loading rate was estimated for pulsed and steady flow

wetlands using weekly hydrologic loading rates and river water nutrient concentrations

(Appendix A, Table A.7). The mean nutrient loading rates to each mesocosm were

-1 -1 similar: 1509 ±349 mg NO3-N wk and 53 ±19 mg total P wk for pulsed wetlands and

-1 -1 1418 ±187 mg NO3-N wk and 57 ±14 mg P wk for steady flow wetlands.

Using the N and P tissue concentrations and plant biomass in August 2003 to compare treatments, pulsed Typha wetlands accumulated the most nitrogen (35.9 ±4.3 g

N m-2, Fig. 2.6) and the most phosphorus (3.13 ±0.18 g P m-2, Fig. 2.7). No significant

differences were detected among hydrological regimes for any of the species groups

(Figs. 2.6 and 2.7).

2.5 Discussion

2.5.1 Plant morphology and primary productivity

The pulsing hydrology did not have a substantial effect on Typha or Schoenoplectus primary productivity when compared to the steady-flow wetlands in this mesocosm study.

Typha had rapid growth during the first two months of the 2003 growing season, accumulating over 60% of its 2003 aboveground peak biomass by mid-June. These observations are consistent with other Typha spp. studies (Martin and Fernandez 1992,

Garver et al. 1988). The range of Typha aboveground dry weight biomass was also consistent with other wetland studies (Dubbe et al. 1988, Mason and Bryant 1975). The root:shoot ratios calculated in this study were within the range (0.9-1.2) found by Kvet and Husak (1978) for T. angustifolia in pond littoral zones in Czechoslavakia. However,

Farnsworth and Meyerson (2003) found that T. angustifolia in Connecticut tidal marshes had lower aboveground biomass (>800 g m-2) but greater root:shoot biomass (2.5 ±0.1).

Schoenoplectus productivity was low compared to literature ranges for natural

wetlands but comparable to other mesocosm studies. In a literature review conducted by

Tanner (2001), peak biomass of S. tabernaemontani ranged widely between 400-1400 g

m-2 for natural marshes in North America. However, in mesocosm studies, the ranges

found in this study are more typical. Ahn and Mitsch (2002) evaluated S.

tabernaemontani productivity in a two-year mesocosm study, maintaining water levels at

10 cm depth and a hydrologic loading rate of 51.1 cm wk-1 and 37.1 cm wk-1 during the

21 two years, respectively. The mean net annual primary productivity of aboveground

Schoenoplectus estimated by peak biomass was 425 ±33 g m-2 (prior to an imposed correction factor). In another two-year mesocosm study conducted by Svengsouk and

Mitsch (2001), mean aboveground Schoenoplectus biomass was greater than 350 g m-2 for control wetlands and between 450 and 600 g m-2 for wetlands fertilized with N, P and both. For their study, groundwater was used and depth was maintained at the soil surface.

The early senescence by Schoenoplectus by mid-August was unexpected. In their review of seed bank and vegetation dynamics in prairie pothole wetlands, van der Valk and Davis (1978) reported that the rhizomes of Schoenoplectus may not tolerate

prolonged high water levels and anoxic/anaerobic conditions. They partially contributed

this condition and the resulting decline of Schoenoplectus to the degenerating vegetation phase of the prairie pothole marsh cycle (van der Valk and Davis 1978). In our study, there was evidence (e.g., the flocculation of Fe along the outflow stand pipes) that soils were highly reduced. However, this condition may have been partially caused by the sunken position of the mesocosm wetlands. It has been demonstrated that mesocosm wetlands can be susceptible to stagnation and less exposed to wind fetch because of their small size and sunken position (Ahn and Mitsch 2002).

2.5.2 Plant tissue nutrient concentrations and uptake

No significant differences between the hydrological regimes were detected for N and

P concentrations separately, but based on N:P ratios, steady-flow wetlands were more N- limited than their pulsed counterparts. It is possible that hydrology affected the availability of both N and P. Steady-flow wetlands remained inundated longer and had

22 the potential to lose more N through denitrification than pulsed wetlands, but based on

Schoenoplectus tissue nutrient data, it appeared that P was the more responsive nutrient to hydrology (Fig. 2.5). The longer periods of anoxic/anaerobic conditions in the steady- flow wetlands may have released more P from the sediment. Evidence of the highly reduced conditions suggests that P (associated with Fe) may have been released during the experimental period, enhancing biomass production. Bayley et al. (1985) found that

merely the presence of standing water, regardless of its nutrient content, can influence

marsh primary productivity and credited this process to the release of P during anoxic

conditions. It is possible that the vegetation in steady-flow wetlands had further uptake of

P relative to N which would explain their slightly lower N:P ratio compared to pulsed

wetlands. While the difference between N:P ratios may or may not be biologically

significant, the statistical difference is noteworthy considering the high N loading rate

during much of the experimental period.

When N:P ratios were examined using the thresholds developed by Koerselman and

Meuleman, (1996) vegetation was considered N limited in both water regimes. These

values are consistent with the results of other N:P ratios from North American marsh

studies cited from the literature by Bedford et al. (1999), although their mean N:P ratio was less (7.5 ±0.7) than either the mean pulsed (11.7 ±0.5) or steady-flow (9.2 ±0.6) ratios calculated in this study. The individual ranges of N and P concentrations and uptake observed in this study are consistent with similar studies. Cronk and Fennessy

(2001) reported Typha spp. leaf tissue concentrations can range from 5-32 mg N g-1 and

1-5 mg P g-1. For Schoenoplectus, it is likely that nutrient translocation to belowground

tissue occurred prior to the onset of senescence and that N and P levels were even higher

23 in July when most Schoenoplectus wetlands were at peak biomass. However, nutrient

proportions between aboveground and belowground plant parts were comparable among

Typha and Schoenoplectus (2-yr) (Figs. 2.6 and 2.7). Tissue concentrations for

Schoenolpectus were also comparable to literature ranges. Tanner (1996) reported that

for second-year Schoenoplectus growth treated with wastewater effluent, N levels were

between 9.2 and 12.9 mg g-1 and P levels between 2.2 and 3.5mg g-1. These N levels are

comparable to those found in this study with the P levels slightly higher.

2.6 Implications for the full-sized wetland pulsing study

The negligible effect of pulsing on most measured parameters was unexpected and

may have been diminished by high hydrologic loading rates which were designed to be

comparable to the full-sized ORWRP wetlands. Even during the later weeks of a pulsed

months when hydrologic input was low, there was still enough water to keep the wetlands

well saturated even though water levels were below the ground surface. This condition

may have reduced some of the prescribed effects associated with pulsing (for instance,

enhanced mineralization though drying and rewetting).

The results of this study suggest that for a full-sized wetlands pulsing experiment as

now underway at the ORWRP, pulsing may only show a negligible effect on Typha and

Schoenoplectus productivity, given the high hydrological loading rate. However, there

are several issues regarding scale that need to be considered before the results of this

study are extrapolated to the full-size wetlands. First, the full-sized wetlands encompass

a large elevation range and will provide a greater range of inundated conditions. Second,

even in areas that remain inundated, the greater exposure to wind fetch and water

24 movement should alleviate the highly reduced conditions that may have negatively

affected the Schoenoplectus in this study. Finally, the mesocosms had vegetation and

soils that had only existed for one or two growing seasons while the full-sized wetlands

have established vegetation and soils that have become depleted of Fe and available P and

therefore may be more tied to fluctuating water levels to release organic-bound nutrients.

Nevertheless, this study has demonstrated that Typha was more resilient to prolonged

hydroperiods and may be better equipped physiologically than Schoenoplectus to

withstand anoxic periods that may occur during steady-flow hydrological regimes.

2.7 Acknowledgements

W. Dick and two anonymous reviewers provided helpful comments that improved

this chapter. Funding for this project came from The Ohio State University, School of

Natural Resources and OARDC Grant No. 2002-079 and USDA Grant No. 2002-35102-

13518. Publication number 05-009 of the Olentangy River Wetland Research Park.

2.8 Literature cited

Ahn, C. and W.J. Mitsch. 2002. Scaling considerations of mesocosm wetlands in simulating large freshwater marshes. Ecological Engineering 18:327-342.

AOAC. 1989. Official Methods of Analysis. Method 990.03. Protein(crude) in Animal Feed Combustion Method(Dumas method). 17th edition 2002. Reference: JAOAC 72, 770.

Bayley, S. E., J. Zoltek, Jr., A. J. Hermann, T. J. Dolan and L. Tortora. 1985. Experimental manipulation of nutrients and water in a freshwater marsh: effects on

25 biomass, decomposition, and nutrient accumulation. Limnology and Oceanography 30:500-512.

Bedford, B. L., M. R. Walbridge and A. Aldous. 1999. Patterns in nutrient availability and plant diversity of temperate North American wetlands. Ecology 80:2151-2169.

Brown, S. L. 1981. A comparison of the structure, primary productivity, and transpiration of cypress ecosystems in Florida. Ecological Monographs 51:405-415.

Casanova, M. T. and M. A. Brock. 2000. How do depth, duration and frequency of flooding influence the establishment of wetland plant communities? Plant Ecology 147:237-250.

Chadde, S. W. 2002. A Great Lakes Wetland Flora.- A complete guide to the aquatic and wetland plants of the upper Midwest. 2nd edition. PocketFlora Press, Laurium, MI, USA.

Cooperrider, T. S., A. W. Cusick and J. T. Kartesz (eds.). 2001. Seventh Catalog of the Vascular Plants of Ohio. Ohio State University Press. Columbus, Ohio.

Cronk, J. K. and M. S. Fennessy. 2001. Wetland plants: biology and ecology. Lewis Publishers, Boca Raton, FL, USA.

Crawford, R. M. M. 1992. Oxygen availability as an ecological limit to plant distribution. Advances in Ecological Research 23:93-185.

Dubbe, D. R, E. G. Graver, and D. C. Pratt. 1988. Production of cattail (Typha spp.) biomass in Minnesota, USA. Biomass 17:79-104.

Fennessy, M. S., J. K. Cronk and W. J. Mitsch. Macrophyte productivity and community development in created wetlands under experimental hydrological conditions. Ecological Engineering 3: 469-484.

Farnsworth, E. J. and L. A. Meyerson. 2003. Comparative ecophysiology of four wetland plant species along a continuum of invasiveness. Wetlands 23:750-762.

Garver, E. G., D. R. Dubbe, and D. C. Pratt. 1988. Seasonal patterns in accumulation and partitioning of biomass and macronutrients in Typha spp. Aquatic Botany 32:115- 127.

Gaudet, C. L. and P. A. Keddy. 1995. Competitive performance and species distribution in shoreline plant communities: a competitive approach. Ecology 76:280-291.

Giovannini, S. G. T. and D. M. L. Da Motta Marques. 1998. Establishment of three emergent macrophytes under different water regimes. Water Science & Technology 40:233-240. 26

Grace, J. B. and J. S. Harrison. 1986. The biology of Canadian Weeds. 73. Typha latifolia L. , Typha angustifolia L., and Typha xglauca Godr. Canadian Journal of Plant Science 66:361-379.

Grace, J. B. and R. G. Wetzel. 1981. Habitat partitioning and competitive displacement in cattails (Typha): experimental field studies. The American Naturalist 118:463-474.

Isaac, R.A. and W.A. Johnson. 1985. Elemental Analysis of Plant Tissue by Plasma Emission Spectroscopy: Collaborative Study. J. Assoc. Off. Anal. Chem.(Vol.68, No. 3. P499, 1985).

Kellogg, C. H., S. D. Bridgham, and S. A. Leicht. 2003. Effects of water level, shade and time on germination and growth of freshwater marsh plants along a simulated successional gradient. Journal of Ecology 91:274-282.

Koerselman, W. and A. F. M. Meuleman. 1996. The vegetation N:P ratio: a new tool to detect the nature of nutrient limitation. The Journal of Applied Ecology 33:1441- 1450.

Kvet, J. and S. Husak. 1978. Primary data on biomass and production estimates in typical stands offishpond littoral plant communities. In: Pond Littoral Ecosystems, D. Dykyjova and J. Kvet (eds.) Springer-Verlag, Berlin. pp.211-216.

Martin, I and J. Fernandez. 1992. Nutrient dynamics and growth of a cattail crop (Typha latifolia L.) developed in an effluent with high eutrophic potential- application to wastewater purification systems. Bioresource Technology 42:7-12.

Mason, C. F. and R. J. Bryant. 1975. Production, nutrient content and decomposition of Phragmites communis Trin. and Typha angustifolia L. Journal of Ecology 63:71-95.

Mcloda, N. A. and R. J. Parkinson. 1980. Soil survey of Franklin County, Ohio. USDA- SCS. US Government Printing Office, Washington, DC, USA.

Mitsch, W.J. and K.C. Ewel. 1979. Comparative biomass and growth of cypress in Florida wetlands. American Midland Naturalist 101:417-426.

27 Mitsch, W. J., J. R. Taylor and K. B. Benson. 1991. Estimating primary productivity of forested wetland communities in different hydrologic landscapes. Landscape Ecology 5:75-92.

Mitsch, W. J., C. J. Anderson, M. E. Hernandez and L. Zhang. 2003. Net primary productivity of macrophyte communities after nine growing seasons in experimental planted and unplanted marshes. In: The Olentangy River Wetland Research Park Annual Report 2002, W. J. Mitsch, L. Zhang and C. J. Anderson (eds.). The Ohio State University, OH, USA. pp 31-36.

Motivans, K. and S. Apfelbaum. 1987. Element stewardship abstract for Typha spp., North American cattails. The Nature Conservancy, Arlington, VA, USA.

Newman, S., J. Schuette, J. B. Grace, K. Rutchey, T. Fontaine, K. R. Reddy, and M. Pietrucha. 1998. Factors influencing cattail abundance in the northern Everglades. Aquatic Botany 60:265-280.

Prentki, R. T., T. D. Gustafson, and M. S. Adams. 1978. Nutrient movements in lakeshore marshes, p. 169-194. In: R. E. Good, D. F. Whigham, and R. L. Simpson (eds.). Freshwater wetlands ecological processes and management potential. Academic Press, New York, NY, USA.

Spink, A., R. E. Sparks, M. Van Oorschot and J. T. A. Verhoeven. 1998. Nutrient dynamics of large river floodplains. Regulated Rivers: Research & Management 14:203-216.

Svengsouk, L.M. and W. J. Mitsch. 2001. Dynamics of mixtures of Typha latifolia and Schoenoplectus tabernaemontani in nutrient-enrichment wetland experiments. American Midland Naturalist 145:309-324.

Svensgouk, L. 1998. First-year response of Typha latifolia L. and Schoenoplectus tabernaemontani (C.C. Gmel.) Palla to nitrogen and phosphorus additions in experimental mesocosms. M.S. Thesis, The Ohio State University, Columbus, OH, USA. 128pp.

Tanner, C. C., J. D’Eugenio, G. B. McBride, J. P. S. Sukias and K. Thompson. 1999. Effect of water level fluctuation on nitrogen removal from constructed wetland mesocosms.

Tanner, C. C. 2001. Growth and nutrient dynamics of soft-stem bulrush in constructed wetlands treating nutrient-rich wastewaters. Wetlands Ecology and Management 9:49-73.

U.S. Environmental Protection Agency. 1983. Methods for chemical analysis of water and wastes. 600/4-79-020, U.S. Environmental Protection Agency, Cincinnati, OH, USA. 28 van der Valk, A. G. and C. B. Davis. 1978. The role of seed banks in the vegetation dynamics of prairie glacial marshes. Ecology 59: 322-335.

Waters, I. and J. M. Shay. 1992. Effect of water depth on population parameters of a Typha glauca stand. Canadian Journal of Botany 70:349-351.

Outer pipe w/protective cap Adjustable stand pipes 0.77 m

Soil 0.60 m 0.30 m

0.10 m Gravel 0.60 m French drain

b) 5.0 cm Soil surface Bleed down orifice (2.0 mm diam.)

Figure 2.1 Experimental wetland mesocosms used in this study. Cross-section view and dimensions of two mesocosms and associated French drain system. Water depth was controlled by the height of the connected stand pipe extending from the mesocosm tub. Bleed-down orifices were installed at the soil elevation of each mesocosm soil elevation.

120.00 Pulsed 100.00 Steady-flow )

-1 80.00

60.00

40.00

20.00

Hydrologic loading wk (cm 0.00 Jul Jul 7 - 13 Jun 1 - Jun 7 Jun 1 - Jun Jun 30 - Jul6 Jun 30 - Jul Jul 14 - 20 Jul Jul 21 - 27 Jul Aug 28 - 3 Jun Jun 8 - 14 May 1 May - 7 Apr 7 - Apr 14 Mar 31 - Apr 6 Aug Aug 4 - 10 Jun 15 - Jun 21 Jun 15 - Jun 29 Jun 22 - Jun May 8 May - 14 Apr 15 - Apr 22 Apr 15 - Apr 30 Apr 23 - Apr Aug 11 - AugAug 18 - 11 May 15 - May 21 May 15 - May 30 May 22 - May

2003 Week

Figure 2.2 Weekly hydrologic loading rate of pulsed and steady flow mesocosms. Total loading input includes pumped river water and rainfall. Pulsed and steady-flow mesocosms received 698 and 694 cm of river water, respectively, during the 2003 experimental period (April-mid July). Arrows indicate the weeks in which pulsing occurred.

31 1800 Below ground biomass )

-2 Above ground biomass- senescent/dead 1500 Above ground biomass- live

Steady- 1200 Pulsed flow

900 Steady- Pulsed flow

600 Steady- Pulsed flow

300

0 ) Above ground dry weight biomass (g m (g weight biomass dry Above ground ) -2 300

600

900

1200

1500 Below ground dry weight biomass (g m (g weightBelow biomass dry ground

1800 Schoenoplectus (1-yr) Schoenoplectus (2-yr) Typha

Figure 2.3 Mean (± 1 SE) total aboveground and belowground biomass (g m-2) for Schoenoplectus (1-yr), Schoenoplectus (2-yr) and Typha mesocosms at August 2003. Aboveground standard error represents total (live and dead) biomass.

Typha Pulsed Pulsed

) Typha 180 Steady 120 Steady 150 -2 100 120 80 90 60 60 40 30 20 No. of ramets m No. 0 0 Mean ramet height (cm Mean ramet height Sept 02 Jun 03 Jul 03 Aug 03 Sept 02 Jun 03 Jul 03 Aug 03

Schoenoplectus (2-yr) Pulsed Pulsed

) Schoenoplectus (2-yr) Steady Steady 180 500

150 -2 400 120 300 90 60 200 30 100 No. of stems m 0 0 Mean stem length (cm Mean length stem Sept 02 Jun 03 Jul 03 Aug 03 Sept 02 Jun 03 Jul 03 Aug 03

Schoenoplectus (1-yr) Pulsed Schoenoplectus (1-yr) Pulsed ) 180 Steady 500 Steady 150 -2 400 120 300 90 200 60 30 100 No. of stems m stems of No. 0 0 Mean stem length (cm Mean length stem Sept 02 Jun 03 Jul 03 Aug 03 Sept 02 Jun 03 Jul 03 Aug 03

Figure 2.4 Mean (± 1 SE) ramet height (cm) and density of Typha and mean (± 1 se) stem length and density of Schoenoplectus (1-yr) and (2-yr) and for pulsed and steady- flow wetlands in September 2002, June 2003, July 2003 and August 2003.

33 1.8

1.6

1.4 14:1 dry mass) dry -1 1.2 16:1

N limited 0.8

0.6

0.4 Schoenoplectus-pulsed

Tissue Phosphorus (mg P g (mg P Phosphorus Tissue Schoenoplectus-steady 0.2 P limited Typha-pulsed Typha-steady 0 0 5 10 15 20 Tissue Nitrogen (mg N g-1 dry mass)

Figure 2.5 Mean N and P concentrations (±1 SE) of aboveground plant tissue of wetland mesocosms in August 2003. N:P ratios of <14:1 and >16:1 are indications of N and P limitations, respectively (Koerselman and Meuleman 1996).

34 a) Typha Pulsed Steady-flow

Inflow: Inflow: 28.8 g 27.0 g -2 -2 NO3 -N m NO3 -N m Live aboveground Live aboveground biomass: biomass: 14.4 ± 2.0 g N m-2 12.0 ± 1.0 g N m-2

Belowground Belowground biomass: biomass: 21.5 ± 2.1 g N m-2 16.4 ± 2.3 g N m-2

Total: 28.4 ± 3.3 g N m- 2 Total: 35.9 ± 4.3 g N m- 2

b) Schoenoplectus (2-yr)

Inflow: Inflow: 28.8 g Live aboveground 27.0 g Live aboveground -2 -2 NO3 -N m biomass: NO3 -N m biomass: 5.5 g N m-2 4.9 ± 0.5 g N m-2 Senescent aboveground Senescent aboveground -2 biomass: 1.8 g N m-2 biomass: 2.5 ± 1.0 g N m Belowground biomass: Belowground 6.8 g N m-2 biomass: 8.9 ± 0.9 g N m-2 Total: 14.1 g N m-2 Total: 16.3 ± 1.3 g N m- 2

Figure 2.6 Mean N retention (±1 SE) for a) Typha and b) Schoenoplectus (2-yr) wetland mesocosms in August 2003. Inflow of NO3-N is based on river water concentrations and pumping rates recorded during the 2003 experimental period (April – August 2003). No significant differences were detected among hydrological regimes for either of the vegetation groups.

35 a) Typha Pulsed Steady-flow

Inflow: Inflow: 1.0 g 1.1 g -2 -2 P m P m Live aboveground Live aboveground biomass: biomass: 1.2 ± 0.1 g P m-2 1.3 ± 0.1 g P m-2

Belowground Belowground biomass: biomass: 1.9 ± 0.1 g P m-2 1.8 ± 0.2 g P m-2

Total: 3.1 ± 0.2 g P m-2 Total: 3.0 ± 0.4 g P m-2

b) Schoenoplectus (2-yr)

Inflow: Inflow: 1.0 g 1.1 g P m-2 Live aboveground P m-2 Live aboveground biomass: biomass: 0.5 g P m-2 0.5 ± 0.0 g P m-2 Senescent aboveground Senescent aboveground -2 biomass:0.2 g P m biomass:0.3 ± 0.1 g P m-2

Belowground Belowground biomass: biomass: 1.2 g P 1.7 ± 0.2 g P

Total: 1.8 g P m-2

-2 Total: 2.5 ± 0.2 g P m

Figure 2.7 Mean P retention (±1 SE) for a) Typha and b) Schoenoplectus (2-yr) wetland mesocosms in August 2003. Inflow of P is based on river water concentrations and pumping rates recorded during the 2003 experimental period (April – August 2003). No significant differences were detected among hydrological regimes for either of the species groups 36

No. of Species No. of growing Hydrology- mesocosms Planted seasons Monthly Pumping Schedule during Experimental Wet Season 1 Schoenoplectus 2 Pulsed - 100 cm/week during week 1, 20 cm/week during weeks 2, 3 and 4 4 Schoenoplectus 2 Steady flow - 40 cm/week during all weeks 2 Schoenoplectus 1 Pulsed - 100 cm/week during week 1, 20 cm/week during weeks 2, 3 and 4 1 Schoenoplectus 1 Steady flow - 40 cm/week during all weeks 5 Typha 2 Pulsed - 100 cm/week during week 1, 20 cm/week during weeks 2, 3 and 4 4 Typha 2 Steady flow - 40 cm/week during all weeks

Table 2.1 Species and hydrology prescribed for wetlands during the experimental hydrology period (April, May 37 and June) in 2003. Actual hydrology period was extended approximately 2 weeks during pump repair. The use of ‘Schoenoplectus (1-yr)’ mesocosms was necessary due to muskrat damage during winter 2002.

CHAPTER 3

THE INFLUENCE OF HYDROLOGIC RESTORATION ON THE PRODUCTIVITY OF A

BOTTOMLAND FOREST IN CENTRAL OHIO

3.1 Abstract

Change in forest productivity in response to hydrologic restoration was evaluated at a 5.2-

ha bottomland hardwood forest in central Ohio. In June 2000, the bottomland forest was

restored to approximate natural flooding by cutting three breeches in an artificial levee

constructed between the river and the forest (north section) and a fourth breech along the

natural river bank to augment flooding at the south section. Total aboveground net primary

productivity (ANPP) was calculated for the two sections of the forest using estimated forest litterfall and wood production. No significant difference in mean ANPP for the north section

(807 ±86 g m-2 yr-1) and the south section (869 ±56 g m-2 yr-1) was detected; however the

north section was substantially more productivity than a previous ANPP estimate conducted

before restoration. A significant positive relationship was detected between ANPP and the

number of days flooded during the year (October 2003-September 2004) in each plot. Forest

ANPP and wood production were also significantly related to total tree basal area and topographic variability. Tree ring-analysis was used to compare mean basal area increment 38 (BAI) growth 10 years (1991-2000) before the restoration to the 4 years (2001-2004) after the restoration. No immediate shifts in BAI were detected; however based on prevailing trends before and after restoration, canopy trees in the south section showed a noteworthy increase in BAI during 2003 and 2004. This shift in the south section was primarily due to the prevalence of boxelder (Acer negundo L.), the dominant species in this section. Evaluating the 14-yr series of BAI for trees in the bottomland, a significant relationship was detected between the total number of days of high-flood conditions (>154 m3 sec-1) and mean BAI

(cm2 yr-1) based on a two-year flooding history.

3.2 Introduction

Bottomland hardwood forests are considered transitional ecosystems because they are influenced by adjacent rivers or streams and terrestrial land upslope. These forests are often highly productive because of the regular influx of nutrients, material and energy from adjacent waterways (Mitsch and Gosselink 2000). The effects of hydrology on riparian forest productivity have been the subject of several studies (Mitsch and Ewel 1979, Brown and

Peterson 1983, Mitsch and Rust 1984, Taylor 1990, Tockner et al. 2000, Mitsch et al. 1991,

Megonigal et al. 1997, Robertson et al. 2001) and most have concluded that periodic flooding has an important influence on the productivity of these ecosystems. According to the subsidy-stress model (Odum et al. 1979), flooding can be beneficial or detrimental to the productivity of the system, depending upon the frequency, timing and duration of the flood events. The model indicates that for a forest at steady-state, periodic flooding provides a nutrient subsidy and thereby increases overall productivity compared to forests in nearby

39 uplands (that do not benefit from the subsidy) or forests in more frequent standing water that

can become physiologically stressed (Teskey and Hinckley 1977a, Kozlowski 1997). The

benefit of surface water connections from the river to floodplains has been demonstrated

along the Danube River in Austria where Tockner et al. (2000) found that floodplains in this

region have the highest productivity when a connection between the river and the floodplain

alternates between a ‘disconnection phase’ (because of low river water levels) and a

‘seepage/downstream surface connection phase’ where low energy inflows of water occur. In

this study, the floodplain benefited from nutrient subsidies from the river, but water levels

also subsided before long-term anoxic conditions occurred that could potentially stress the

forest. Despite application of the subsidy-stress model in several studies, other studies have

found that the highest productivity occurred in forested regions other than those periodically

flooded. Brown and Peterson (1983) and Burke et al. (1999) found that permanently flooded

zones rather than periodically flooded zones had higher productivity while Megonigal et al.

(1997) found no difference between upland and periodically flooded forest productivity. The

Megonigal et al. study supported an earlier model presented by Mitsch and Rust (1984)

which holds that the potential benefits derived from periodic flooding are offset by the

physiological stress induced by anaerobic soil conditions. Evaluating tree rings of three

floodplain species along the Kankakee River in northeast Illinois, Mitsch and Rust (1984) did

not find a relationship between radial growth and flooding duration but instead attributed tree growth to a combination of hydrologic and climatologic factors that can influence soil moisture.

In most bottomland forests, the larger, canopy-sized trees often provide the majority of forest production (Kimmins 1987); therefore the response of this stratum to changes in

40 hydrology will typically dictate overall forest-level productivity. Numerous tree species can

be present in a bottomland community and it has been shown that different species will have

different responses and tolerances to flooded conditions (Teskey and Hinckley 1977b,

Kozlowski 1997). For instance, Dudek et al. (1998) found different responses to

hydrological cues when comparing the long term growth of a flood tolerant species (Populus

deltiodes Marsh.) and a flood intolerant species (Juglans nigra L.) growing in the same

central Ohio bottomland forest used in this study.

Given the inconsistencies in bottomland responses to flooding, it has been suggested that

more studies need to evaluate existing forests under a changing hydrology to elucidate the

influence of hydrology (Conner 1994, Megonigal et al. 1997). Our study was conducted to

evaluate short-term forest responses to the hydrologic restoration in a bottomland hardwood

forest at the Olentangy River Wetland Research Park (ORWRP) in central Ohio. The

objectives of this study were to determine if: 1) the reconnection of the north section of the

bottomland forest to the adjacent Olentangy River increased aboveground net primary productivity (ANPP) after four years, 2) flood frequency or other ecological conditions within the bottomland could be used to predict ANPP, 3) there has been a response (positive or negative) in the average or species annual radial growth rate of canopy trees since hydrologic restoration, and 4) the frequency of previous flood events could be used to predict tree radial

growth.

3.3 Methods and materials

3.3.1 Study site

The 5.2-ha bottomland hardwood forest at the Olentangy River Wetland Research Park

(ORWRP) is located along the Olentangy River, a 4th order river in central-Ohio USA. The

ORWRP bottomland forest varies between 25-90 m wide, is approximately 730 m long, and

was hydrologically restored starting in June 2000 (Fig. 3.1). Hydrologic restoration was

conducted as partial wetland mitigation by the Ohio Department of Transportation for

wetland impacts associated with a highway project in Columbus, Ohio. The north section of

the bottomland forest was disconnected from river flooding by a constructed levee (up to 2-m

high) that was built over 70 years ago (Cochran 2001) and extended along a 250 m stretch of

the river. Three breeches (Cuts #1-3, Fig. 3.1) were opened in the north section levee and

river water now regularly flows into and out of this section of the bottomland during high

river events. The levee only affected the north section of the bottomland. The south section

of the bottomland was not restricted by artificial levees and periodically flooded by direct

surface flow from the river; however floods were infrequent and only occurred during extremely high river events. To increase flood frequency and create flow-through conditions,

a fourth breech (Cut #4) was made through the natural river bank to a lateral swale which

extends through the south section (Fig. 3.1).

In a previous study of riparian forest productivity in this bottomland hardwood forest, forest productivity and the basal growth of canopy trees (>25cm dbh) were evaluated using data collected between 1998 and 2000 (Cochran 2001). Mean ANPP of the ORWRP bottomland (averaged between sections) was estimated at 800 g m-2yr-1, substantially lower

42 than productivity of two other unrestricted bottomlands upriver that averaged 1280 g m-2yr-1.

Higher productivity in the unrestricted bottomland forests was attributed to their ability to receive river influx and higher proportion of species adapted to these conditions.

3.3.2 Climate and hydrology

River stage has been measured twice nearly every day since 1994 using a permanent staff gauge immediately upriver from the ORWRP bottomland (Fig. 3.1). When water levels were high enough to flood portions of the bottomland, we observed the spatial extent of flooding within the forest relative to river stage and recorded observations in river inflow sources

(Cuts #1-4), internal flow patterns, and relative depths at various river stages. Precipitation and weather data were gathered from a Columbus, Ohio weather station operated by the Ohio

Agricultural Research & Development Center (www.oardc.ohio-state.edu/centernet/ weather.htm).

3.3.3 Aboveground net primary productivity

To determine the effect of the restored hydrology on bottomland productivity, wood and litterfall production data were collected to determine annual aboveground net primary productivity (ANPP) (Newbould 1967). A transect was established within the north and south sections of the forest. Transects were randomly established but designed to extend parallel to the river and through the regularly flooded portions of both sections. Because of the wider forest in the south section, parallel transects were used to increase plot replication.

A total of 10 plots (20 m x 25 m) were measured and marked in the field (Fig. 3.2).

43 In each plot, all trees with a dbh (1.3m) >5cm were identified by species, tagged and

measured for dbh in April 2004 and April 2005 to determine 1-yr basal increase. Using tree

data, species importance values were calculated in 2004 using the following equation:

Importance value = relative density + relative dominance + relative frequency (1)

2 -1 The increase in tree basal area (Ai) (cm yr ) was calculated by the following equation

(Newbould 1967):

2 2 Ai = π [r -(r-i) ] (2)

Where, r = radius of tree at breast height (cm), and

i = radial increment per year (cm2 yr-1)

Tree heights were measured using a clinometer in May 2005 and the annual wood

-1 production per tree (Pi)(g yr ) was calculated by the following parabolic volume equation

(Whittaker and Woodwell 1968, Phipps 1979):

Pi = 0.5ρ Ai h (3)

Where, ρ = wood specific gravity (g cm-3), and

h = tree height (m)

44 Wood specific gravity values were obtained from the U. S. Forest Products Laboratory

(1974) and Alden (1995). The plot wood production was calculated as the summation of all

wood production per tree and converted to g m-2 yr-1.

A total of 50 leaf litter traps (5 per plot) were installed in May 2004. Each plot was

divided into 4 quadrants and a leaf trap was randomly placed in each quadrant with a fifth

trap randomly placed near the center (Fig. 3.2). Leaf traps were 15 cm tall, 0.25 m2 in area,

lined with 2-mm screen and installed approximately 1.0 m off the ground to avoid flooding and litter saturation. Litterfall was collected for one year starting in May 2004. Traps were

emptied twice a month from June-December and once a month from January-May. After

each collection, the contents were separated into leaves, reproductive material and woody

material, air-dried at room temperature for 1 week and then at 105ûC for four days or until constant mass prior to being weighed. Leaf traps were averaged per plot and the summation of all fine litter production (leaf litter and reproductive materials) was calculated. Because of vandalism and flood/ice damage, several sampling periods had plots with less than the 5 traps available and were averaged only using the plots that were undamaged.

Using litterfall and wood production data, aboveground net primary productivity (ANPP)

(g m-2 yr-1) for each plot was estimated using the following equation (Whittaker and

Woodwell 1968):

ANPP = plot wood production + litterfall production (4)

45 3.3.4 Predicting ANPP, litterfall production and wood production

Various environmental parameters known to influence forest productivity were selected

to predict forest productivity in 2004 (ANPP, wood production and litterfall) through linear

regression. The 2004 river hydrograph and observations of flooded conditions at different

river stages were used to determine 1) the number of flood events that directly connected to

each plot, and 2) the number of days that the river had a surface water connection to each

plot. Flooding frequencies and durations in 2004 for each plot were estimated for the

preceding year (October 2003-September 2004), preceding two years (October 2002-

Sepetmebr 2004) and the growing season (April-September 2004) and used for regression

analyses.

To assess the potential influence of tree plot elevation on ANPP, the corners of each plot

and each random leaf litter trap within the plot quadrants (Fig. 3.2) were surveyed for

elevation using a TOPCON RL-H3CTM rotating laser level and the mean plot elevation (m

MSL) was calculated. To assess the potential influence of topographic variability on forest

productivity, the variance of all elevation points at each plot was also calculated and used to

predict forest productivity.

Other data used as predictor variables included canopy cover and tree basal area. Canopy

cover (%) was estimated for each plot in August 2004 using a convex spherical crown

densitometer. Cover was measured at each trap facing the four cardinal directions and the

mean of all measurements were calculated for the entire plot. Tree basal area (cm2 m-2) per plot was calculated based on the total basal area of all trees >5 cm dbh measured in April

2004.

46 3.3.5 Tree-ring analysis

For each forest canopy tree (>25 cm dbh and >15 m height) in the plots, two cores were

extracted using a 5.15 mm inside increment borer. Seven supplemental trees located between

tree plots (5 in the north section and 2 in the south section) were added. For comparison with trees not in the flood zone (the upland area between sections) a total 7 trees from species representative of the flooded sections were randomly selected for coring. For each tree, two

cores were taken at 90° angles from each other to account for natural variation and were

collected at least 12 cm into the tree to collect >15 years of increment growth. The cores

were temporarily stored in straws, air-dried and then glued into grooved holders. Cores were

sanded with a series of finer sandpaper grit (80-600) and polished with lamb’s wool. Tree

cores were scanned and the image was analyzed for tree-ring widths (to the nearest 0.01 mm) using WinDENDRO TM (2002). Replicate tree-ring increments were compared for

comparable growth patterns, verified with a stereoscope when necessary and averaged for

each tree.

Using the tree cores and tree diameter, basal area (Ai) increments (BAI) were calculated

from 1991 to 2004. Years 1991-2000 were selected as representative pre-restoration growth

and years 2001-2004 were analyzed as post-restoration years. Although most restoration

work was conducted in June 2000, each cut was excavated further in early 2001. The first

flood event to overflow into bottomland forest did not occur until April 2001. For

comparison of trees between sections and species, the BAI (cm2 yr-1) from each tree were

standardized to reflect percent basal increase relative to total tree basal area [BAI (%)], and

were calculated using the following equation:

47 BAI(%) = [(Ai Year X – Ai Year X-1)/Ai Year X-1] * 100 (5)

3.3.6 Predicting basal growth

The series of BAI data collected for 1991-2004 (both cm2 and %) were evaluated to

determine if flood stage (based on river discharge) could be used to predict basal growth. A

discharge curve prepared for river depth at this section of the Olentangy River (Mitsch 1995)

was used to determine the number of bank-full flood days/events (221.2 mMSL or >70 m3

sec-1) and the number high-flood days/events (221.6 mMSL or >154 m3 sec-1) between March

1994 and September 2004. The high-flood discharge was selected because this is the

discharge level that was likely required to directly flood both sections of the bottomland

despite the presence of the levee. Daily river discharge data from an upstream United States

Geological Survey (USGS) stream gauge (near Delaware, Ohio, Station No. 03225500) was

used to estimate discharge rates at the study site between October 1990 and March 1994. A

regression between known ORWRP and USGS discharge rates were used to estimate

discharge on the days where no water level data was available at the study site (ORWRP=

1.43*USGS + 5.34, R2=0.92).

Using daily river discharge data, the frequency and duration of flood events were

determined for each year (from 1 October in the preceding year to 30 September) and

growing season (1 April to 30 September). Frequency and duration were determined for

bank-full and high-flood discharge events. For both thresholds, the number of days and

events in which these rates occurred were counted for each applicable year and growing

season.

48 In addition to conducting a regression analysis on the concurrent flood and BAI data for a given year, regressions were also conducted to evaluate the possibility of a lag in tree basal growth response to floods. A regression analysis was used to evaluate flood frequency and

BAI data lumped into 2-yr increments. In addition to capturing potential lag effects, lumping

BAI data in this manner has been suggested as an effective guard against potential errors due to false-rings or other measurement errors (Mitsch et al. 1991). A second regression analysis was conducted using two years of preceding river discharge data to predict the BAI for a given year.

3.3.7 Statistical analyses

An independent t-test was conducted to compare the mean ANPP between the north and south sections of the bottomland. Because litterfall and wood production have been shown to respond independently to environmental factors, independent t-tests were also conducted to compare these parameters. Analyzing tree-ring data, paired t-tests and trend analyses were used to compare BAI (%) between pre- and post-restoration years for each section. Similarly, paired t-tests and trend analyses were used to compare BAI (%) for pre- and post restoration specimens of A. negundo and A. glabra. All pre- and post-restoration data were tested for normality using the Kolmogrov-Smirnov test, homogeneity of variances using Levene’s test, and transformed as needed to meet test assumptions. For all t-tests, p-values <0.05 were considered significant differences and p-values <0.01 were considered highly significant.

Regression analysis was used to evaluate relationships between forest productivity

(ANPP, litterfall production and wood production) and measured environmental variables

[flooding frequency (total year, total year + preceding year, and growing season), flooding

49 duration (total year, total year + preceding year, and growing season), elevation, topographic variability, total tree basal area and canopy cover] at each plot. Best-fit regression analysis was conducted to determine the most appropriate model type (linear or polynomial).

Significance of the regression analyses were tested with analysis of variance with p-values

<0.05 considered a significant and p-values <0.01 considered highly significant. All response and predictor variables were tested for normality using the Kolmogrov-Smirnov test and homogeneity of variances using Levene’s test. Variables not meeting test assumptions were transformed as needed. Where the regression of time series data was conducted, an autocorrelation function (1- or 2-year lags), the Durbin Watson test, or both were conducted as appropriate. Minitab™ v.14 was used to run all statistical analyses.

3.4 Results

3.4.1 Hydrology and climate

Based on precipitation data and a hydrograph of the Olentangy River (Figs. 3.3 and 3.4), conditions in the post-restoration period tended to be wetter than normal. Between 1991 and

2000, there were only two years (1995 and 1996) where precipitation was exceptionally high

(>20 cm above normal for any 3-month period) and one year (1999) where is it was exceptionally low. In contrast, three out of four of the post-restoration years (2002-2004) had exceptionally wet seasons. However, these wet seasons were offset by drier than normal winter seasons. Nevertheless, frequent high river levels were common during those years.

As indicated on the post-restoration river hydrograph (Fig. 3.4), river levels frequently met or exceeded the designed bottomland flood level (221.2 m MSL) from 2002-2004 compared to much less frequently in 2001. It was noted during this period that Plot #5 in the south section

50 (Fig. 3.2) was too high in elevation to become regularly flooded (unlike all the other plots)

and therefore it was removed as part of the south section and analyzed separately as an

upland plot.

Floods tended to be short-term events and rarely lasted more than a few days. Flood

waters tended to rapidly rise and then fall back to normal flow levels (220.6 m MSL). Length

of inundation after flooding occurred was not systematically measured, however it was

normal for water to rapidly drain from low spots in the bottomland forest after only a few

days, depending upon the flood stage, post-flood river levels and season. Winter flood water

often froze once in the bottomland and may last for weeks while summer flood waters dried

out the quickest (presumably because of enhanced transpiration). Tree plots within the

bottomland connected with the river at different river stages with Plots #1, 6, 8 and 10 being

the first to flood. Consequently, during minor flood events (between 221.2 and 221.4 m

MSL), these plots would connect with the river while the others would not. Approximate

river stage at which each plot was flooded was determined and based on hydrograph data, the

number of days and flood events affecting each plot was determined for the entire 2004 year

and growing season (Table 3.1).

3.4.2 Bottomland composition

Based on the identified trees >5 cm dbh, a total of 386 trees representing 19 species were

accounted for in the bottomland forest plots. A total of 257 of these trees were in the north

section (or 1285 trees ha-1) compared to 129 trees in the south section (or 516 trees ha-1).

Forest composition was different between the two sections with the north section having a higher proportion (62%) of small trees (5-10 cm dbh) compared to the south section (44%).

51 Size distribution was fairly even in the south section with subcanopy trees (10-25 cm dbh)

constituting 25% of the total and canopy trees (>25cm dbh) at 30%. Subcanopy and canopy

trees in the north section constituted 27% and 11%, respectively.

Understory trees in the north section have become dominated by paw paw (Asimina

triloba L.) to the extent that they have become the more dominant species in terms of

importance value (Table 3.2; Appendix B, Table B.1). Other trees with high importance

values in the north section included Ohio buckeye (Aesculus glabra Willd.), hackberry

(Celtis occidentalis Willd.) and boxelder (Acer negundo L.). Trees in the south section were

dominated by overstory species A. negundo and to a lesser extent A. glabra and eastern

cottonwood (Populus deltiodes Bartr. Ex) (Appendix B, Table B.2).

3.4.3 Aboveground net primary productivity

There was no significant differences detected in ANPP between the north section (807

±86 g m-2 yr-1) and the south section (869 ±56 g m-2 yr-1) (Fig. 3.5). Significantly higher (t=-

2.86, df=5, P<0.05) mean litterfall production was detected in the south section (555 ±32 g

m-2 yr-1) compared to the north section (460 ±9 g m-2 yr-1) (Appendix B, Table B.3-5).

Unlike litterfall production, wood production (Appendix B, Table B.6) was highly variable in

the north section ranging between 157-535 g m-2 yr-1. No significant difference was detected

between mean wood production in the north (346 ±82 g m-2 yr-1) and south section (314 ±33

g m-2 yr-1). Productivity in Plot #5 (which was converted to an upland plot based on its

elevation) had ANPP, litterfall and wood production (855, 544 and 311 g m-2 yr-1,

respectively) that was comparable to plots in the adjacent south section.

3.4.4 Predicting ANPP, litterfall production and wood production

Using plot-level flooding and productivity data, a significant relationship between the total number of days flooded in 2004 (October 2003-September 2004) and ANPP was detected (R2=0.44, P=0.050). Furthermore, when flooding based on river levels from the

preceding year were added (2003 and 2004), significant relationships were detected between

ANPP and the total number of flooded days (R2=0.48, P=0.040, Fig. 3.6) and the total number of days flooded in the growing season (R2=0.46, P=0.040). Regression analyses

determined that none of the flood frequency parameters calculated had an influence on the

separate components of ANPP (litterfall or wood production).

Both ANPP and wood production were significantly influenced by plot topographic

variability (elevation variance) (Fig. 3.7; Appendix B, Table B.7) and total tree basal area

(cm2 m-2) (Fig. 3.8). Elevation variance data was log-transformed to meet normality assumptions. No significant relationships were detected between any predictor variables and litterfall production. A synopsis (range, mean and standard error) of all predictor variables used to predict forest productivity through regression analysis is provided in Table

3.1 (also see Appendix B, Table B 6-8).

3.4.5 Tree-ring analysis for BAI

Comparing the mean BAI (%), canopy trees in the north and south section decreased in mean increment size after the restoration (P>0.01 and P=0.03, respectively, Table 3.3;

Appendix B, Table B.9). However, no significant changes were detected in actual annual

BAI (cm2 yr-1), suggesting that trees maintained consistent wood production since 1991 while

53 increasing in age. No significant changes in BAI (% or cm2 yr-1) were detected in upland

trees. It was noted during the analysis that 5 of the largest tree specimens (all >75 cm dbh)

had consistently low BAI (%) values that had an excessive influence on mean comparisons

and trend analyses, and were therefore omitted. We presumed that these older trees had

reached an age where a high proportion of gross production is used for maintenance

metabolism (Kimmins 1987) and were unlikely to provide a growth response to changing

moisture conditions.

Evaluation of trend analyses showed that none of the sections had an abrupt shift in basal

growth immediately after hydrologic restoration (Fig. 3.9). However, canopy trees in the

south section showed increased radial growth in 2003 and 2004 compared to a trend of

consistent decline in BAI(%) since 1994. Trees in the north and upland sections showed a

slight increase of BAI(%) in 2004, however conditions in both sections during pre-restoration

years were more variable making this shift difficult to assess.

Two of the most dominant trees in the bottomland forest (A. negundo and A. glabra) were

evaluated separately to see if responses between species were different. Because similar

trends were detected in A. negundo between the north and south sections, these trees were

pooled. No canopy-sized A. glabra occurred in the north section plots. Like trees in the

north and south sections, A. negundo had a significantly lower mean BAI (%) after the

restoration (P>0.05) but with no significant change in BAI (cm2 yr-1) (Table 3.4). Trend analyses indicated that unlike canopy specimens in the upland section, A. negundo trees in the

flooded sections may have responded positively to the restoration based on the increased BAI

(%) in 2003 and 2004 (Fig. 3.10a). A. negundo trees in the upland sections seemed to follow a basal growth trend that extended back to 1991.

54 The BAI (% and cm2 yr-1) of A. glabra canopy trees were not significantly different

between pre- and post-restoration years (Table 3.4). Upland and flooded specimens had

similar BAI (%) extending back to 1991 (Fig. 3.10b). After 2000, there was a separation between the upland and flooded trees, however BAI trajectories did not shift substantially in

the post-restoration period.

3.4.6 Predicting basal growth

A significant relationship was detected between the total number of days where the river discharged at high-flood stage (>154 m3sec-1) and BAI (cm2 yr-1) when analyzed using the

preceding 2-yr river data (Fig. 3.11a). Similarly, a significant relationship was detected

between the number of high-flood days over a 2-yr period and the corresponding 2-yr BAI

(cm2 yr-1) (Fig. 3.11b). No significant relationships were detected between the number of

days or events of discharge and the BAI for that corresponding single year. A significant

relationship between the total number of high-flood discharge events and 2-yr preceding river

data was also detected (R2=0.54, F=13.93, P=0.003), but the number of events was less

predictive than the number of days. No other significant relationships between BAI and river

discharge were detected. In all cases, BAI (%) data showed indications of autocorrelation

and therefore were omitted from consideration in favor of BAI (cm2 yr-1).

3.5 Discussion

3.5.1 Bottomland productivity

One of the most commonly cited benefits associated with the hydrologic restoration of a

bottomland forest is the likely enhancement in productivity. Based on the results of this

55 study there is some evidence to suggest that after only four years, there was an increase in

bottomland productivity. In terms of ANPP, we found no significant differences between the north (807 ±86 g m-2 yr-1) and south (869 ±56 g m-2 yr-1) sections. This was important because using productivity data from plots comparable to our study, Cochran (2001) found that ANPP in the north section (531-641 g m-2 yr-1) was significantly lower than the south

(793-1033 g m-2 yr-1). This suggests that the north section has increased in productivity since

the restoration activity occurred. The biggest difference between pre- and post-restoration

productivity in the north section was in mean wood productivity which, in 2004 (346 g m-2 yr-1, this study) was nearly triple that estimated in 2000 (117 g m-2 yr-1, Cochran 2001).

However, the change in wood productivity conflicts somewhat with our tree canopy ring-

analysis data which saw relatively consistent basal area growth (cm2 yr-1) between pre- and post-restoration years in the north section. Given the high variability of wood production estimated for plots in the north section, plot location may have greatly influenced estimates in both studies. Cochran (2001) only used 2 plots (20 x 25m) in the north section directly affected by the levee, compared to 4 plots used in this study. Therefore we conclude only tentatively that ANPP has increased in the north section.

Based on estimates by Cochran (2001), ANPP in the north section had clearly exceeded its pre-restoration range while in the south section ANPP was still within the pre-restoration range. Furthermore, the ANPP range seen at the bottomland forest was still below what has been recorded at other sections of the Olentangy River. At two other unrestricted bottomland hardwood forests upriver from the ORWRP (both within 12 km), forest ANPP was estimated at 1283 ±56 and 1297 ±302 g m-2 yr-1 (Cochran 2001). The ANPP range seen at the ORWRP

bottomland also seems to be lower than what has been observed at most other bottomland

56 forests in the region. Mitsch et al. (1991) found ANPP at 1280 and 1334 g m-2 yr-1 in two

hardwood bottomland forests along the Ohio River in western Kentucky. ANPP for a

floodplain forest in Illinois was estimated at 1250 g m-2 yr-1 (Johnson and Bell 1976).

However, Brown and Peterson (1983) found that ANPP at another bottomland forest in

Illinois with stagnant water conditions was 960 g m-2 yr-1 while a seasonally flooded forest

was at 668 g m-2 yr-1. It seems that in terms of long-term productivity, the ORWRP

bottomland may still have an opportunity to increase.

Although leaf productivity was significantly higher at the south section, it appeared that wood production was more the responsive component affecting ANPP based on the wide ranges observed at the ORWRP (Fig. 3.5). This in contrast to other studies (Burke et al.

1999) which found leaf production to be more variable. Part of the reason that litterfall was

more consistent between plots may have been the frequent occurrence of paw paw (A.

triloba) in north section plots. Although these plots had less canopy-tree cover and overall

basal area, there was a large contribution of litterfall provided by subcanopy A. triloba which

produced a dense cover of large leaves.

3.5.2 Relationship between bottomland productivity and flooding

Although plot ANPP was predicted by the number of days each was flooded in 2004, the

best relationships were found using flood data added from 2003 and 2004. The results of

these analyses confirmed that surface water flooding was an important factor in determining

forest productivity and also suggests that flood events may influence productivity beyond the

year they occur. Similar patterns were revealed using river discharge to predict basal tree

growth (see Section 3.5.4 below). It is possible that this delayed response represents the

57 time it takes for deposited nutrients to desorb from sediment and mineralize from matter and become available. The decomposition of organic matter, the desorption of nutrients from sediment and the alteration of soil chemistry are all factors that dictate nutrient availability in bottomland soils (Mitsch and Gosselink 2000). The rates of these processes are eventually dependent upon environmental conditions including hydrology and climate. Therefore if it takes several months for ecological processes to make nutrients available, nutrients from material deposited in the spring and early summer (when most flooding traditionally happens) may not become available to plants until the subsequent growing season.

Using regression analyses, total ANPP and wood production were significantly influenced by total basal area and topographic variability (elevation variance). It was no surprise that existing basal area influenced productivity however elevation variance was one of the least considered predictor variables at the onset of this study. Floodplain bottomlands can have naturally diverse topographies consisting of repeated ridges, swales and meandering scrolls

(Leopold et al. 1964). The influence of topography has been demonstrated on forest productivity in the southern Appalachian (Bolstad et al. 2001), on riparian plant diversity in

Alaska (Pollock et al. 1998) and canopy gap regimes in a Texas bottomland forest (Almquist et al. 2002), however there is little information pertaining to its influence on bottomland tree productivity. A diverse topography such as that of a ridge-and-swale would perhaps allow the greatest interface between flood waters and trees on slightly elevated ground. In the case of the ORWRP bottomland, topographic variability was provided by swales and ridges in the south section, however in the north section it was provided by the old fill material from the remnant levee. The influence of topography on bottomland productivity is an interesting

58 result from this study and we would encourage future bottomland research to consider this component.

3.5.3 Forest basal growth before and after restoration

Evaluating canopy tree cores, we did not find an occasion where radial tree growth made an immediate and clear response to the restored hydrology. Given that the north section was

a more complete restoration (hydrology was only enhanced in the south section) we were

expecting to see a positive response to the restored hydrology. However, compared to the

other sections, only the south section showed a potential response. The change in BAI (%)

seen at the south section in 2003 and 2004 was interpreted to be a more significant shift

because it represented a clear break in a very consistent growth trend dating back to 1994.

An increase was detected in the north section in 2004, however given the modest size of the

increase, the more sporadic growth trend leading up to it, and that upland trees showed a

similar increase; this change cannot be considered conclusive. It may be that because the

south section trees were exposed to occasional flooding prior to the restoration work, trees in

this section were better conditioned to the altered hydrology. Assuming that the increased

flooding has been a stress to trees, when stressors are introduced more gradually, trees can

generally make the physiological adjustments to protect themselves much more than if the

stressor is introduced rapidly (Kozlowski and Pallardy 2002). The canopy tree response in

the south section may have been in response to the high inflows that occurred in 2003 and

2004, or perhaps more likely, it may be a lag response to the new hydrology. This would not

be unprecedented, as lags in forest response have been documented in the case of other

habitat improvements. Jones and Thomas (2004) found that in Ontario forest stands

59 dominated by sugar maple (Acer saccharum Marsh), peak growth enhancement in response

to canopy gaps did not occur until 3-5 years later. Given the shift in hydrology is even more

substantial in the north section it may take longer for trees there to positively respond.

Anaerobic conditions caused by flooding may have been exacerbated in this section where flooding was previously rare.

A. negundo was the dominant tree in the south section and therefore its trend in BAI (%) over time (Fig. 3.10a) was similar to that seen for all south section trees (Fig. 3.9). However,

A. negundo specimens tended to respond similarly in the north section as well. The response of trees in the flooded sections since the restoration is in contrast to upland specimens where

BAI (%) maintained the same trend set before the restoration occurred. The physiology of A. negundo may make it well adapted to changing water conditions as it has been shown that its net photosynthesis can be resilient to seasonal changes in soil water potential (Foster 1992).

A. glabra on the other hand did not show a substantial response although its BAI (%) has not declined during the post-restoration period as the upland specimens have. Nevertheless, this tree tends to occurs in moist soils and while it is considered resistant to saturation, it is a facultative upland species and might be less resilient to prolonged anaerobic conditions.

3.5.4 Basal growth in response to flooding

Based on the results of this study, there is evidence that flooding may have a lagged effect on tree growth. In both scenarios where river discharges from the current and previous years were added, there was a significant relationship between the number of days with high-flood discharge and BAI (cm2 yr-1). Given the pre-restoration exclusion of bank-full flood waters it

is not surprising a relationship was only detected using the high-flood events, and as

60 indicated in Figs. 3.11a and b, the bottomland forest was still responding to these high-flood

occurrences during the post-restoration years. The evidence of a lagged response by

bottomland canopy trees to flooding has been rarely documented however it isn’t unexpected given the amount of other circumstances where forests have shown a lagged response in

growth. Factors such as climate (Fritts 1976, Camill and Clark 2000), newly formed canopy

gaps (Jones and Thomas 2004) and the removal of shelterwoods (Holgen et al. 2003) have all

been shown to induce a lagged response on tree basal growth.

Significant regressions using current- and previous-year river discharge data indicated

that basal tree growth occurred at an optimal number of high-water discharge days (~10)

suggesting that trees are benefiting from a nutrient subsidy to a point. After about 10 high-

flood discharge days, the bottomland may no longer be nutrient limited and anaerobic

conditions may have reduced productivity. It is important to point out that in the 2-yr periods

where high-flood discharge exceeded 10 days, the decrease in basal growth was marginal

compared to those years where floods events were scarcer. The general relationship seen in

this case is not unprecedented. Golet et al. (1993) showed that the highest tree basal growth

at red maple (Acer rubrum L.) swamps in Rhode Island occurred at intermediate annual water

levels. The results from this study support findings such as these and demonstrate the push-

pull influence that flooding has on forest productivity.

The fact that flooding throughout entire years (and not just the growing seasons) was the

best predictor of basal growth supports the idea that these trees were responding more to a

nutrient subsidy and less to the anaerobic stress of flooding. If flooding stress was more

important, we would have expected a relationship with BAI to manifest during the growing

season. However, as seen in other studies, it is likely that the anaerobic stress caused by

61 flooding in the growing season was negated by a nutrient subsidy, and therefore a relationship between growing season flood occurrence and BAI was unapparent. Furthermore, it appears that trees are responding to sediment and nutrient deposition occurring year-round. Through the work of Zhang et al. (2005) and personal observations, it has been shown that these flood events can deposit significant amounts of material into the bottomland forest and the amount of material, sediment and nutrients available to trees may ultimately be dependent upon the frequency of major flood events in the preceding years.

3.6 Conclusions

Hydrologic restoration of the ORWRP bottomland forest was conducted in 2000 and as a result, the north section has received direct surface flows from river floods and the south section has increased its surface flow and frequency. The two sections were similar in

ANPP, but compared to previous estimates conducted before the restoration, the north section has increased its mean ANPP since the restoration occurred. No abrupt and clear changes in canopy tree basal growth has occurred since the restoration occurred, however since 2003, trees in the south section of the bottomland have shifted from a continuous trend of declining

BAI (%) extending back about ten years. Evaluating BAI and river discharge data since

1991, these results suggest that for a two-yr period, optimal basal growth will occur when

~10 days of high-flood discharge occur during that period. These results also suggest that basal growth in response to flooding is lagged by at least one year as no relationships were detected between tree basal growth and concurrent flooding over one year. The lack of any significant relationship between tree basal growth and flooding in the growing season

62 suggests that sediment and nutrient deposition are likely more important to forest productivity than the stress caused through flooding.

3.7 Acknowledgements

This study was partially funded through a contract with the Ohio Department of

Transportation. Field assistance was provided by Jeremiah Miller. Editorial comments by

Charles Goebel improved this chapter.

3.8 Literature cited

Alden, H. A. 1995. Hardwoods of North America. U. S. Department of Agriculture, Forest Service, Madison, WI, USA. FPL-GTR-83.

Almquist, E. B., S. B. Jack and M. G. Messina. 2002. Variation of the treefall gap regime in a bottomland hardwood forest: relationships with microtopograpy. Forest Ecology and Management 157:153-163.

Baker, III, T. T., W. H. Conner, B. G. Lockaby, J. A. Stuart, and M. K. Burke. 2001. Fine root productivity and dynamics on a forested floodplain in South Carolina. Soil Science Society of America Journal 65:545-556.

Bolstad, P. V., J. M. Vose, and S. G. McNulty. 2001. Forest productivity, leaf area, and terrain in southern Appalachian deciduous forests. Forest Science 47:419-427.

Brown S. and D.L. Peterson. 1983. Structural characteristics and biomass production of two Illinois bottomland forests. American Midland Naturalist 110:107-117.

Camill, P. and J. S. Clark. 2000. Long-term perspectives on lagged ecosystem responses to climate change: permafrost in boreal peatlands and the grassland/woodland boundary. Ecosystems 3:534-544.

63 Cochran, M. 2001. Effect of Hydrology on bottomland hardwood forest productivity in central Ohio (USA). M.S. Thesis. The Ohio State University, Columbus, OH, USA.

Conner, W. H. 1984. Effect of forest management practices on southern forested wetland productivity. Wetlands 14: 27-40.

Conner, W. H. and J. W. Day, Jr. 1982. The ecology offorested wetlands in the southeastern United States. p. 69-87. In B. Gopal, R. E. Turner, R. G. Wetzel, and D. F. Whigham, (eds) Wetlands: ecology and management. National Institute of Ecology and International Scientific Publications, Jaipur, India.

Dudek, D.M., J.R. McClenahen and W.J. Mitsch. 1998. Tree growth responses of populus deltoides and Juglans nigra to streamflow and climate in a bottomland hardwood forest in central Ohio. American Midland Naturalist 140:233-244.

Foster, J. R. 1992. Photosynthesis and water relations of the floodplain tree, boxelder (Acer negundo L.). Tree Physiol. 11:133-149.

Golet, F. C., A. J. K. Calhoun, W. R. DeRagon, D. J. Lowry, and A. J. Gold. 1993. Ecology of red maple swamps in the glaciated northeast: a community profile. Biological Report 12, U.S. Fish & Wildlife Service, Washington, DC. 151 pp.

Holgen P., U. Soderberg, and B. Hanell. 2003. Diameter increment in Picea abies shelterwood stands in northern Sweden. Journal of Forest Research 18:163-167.

Johnson, F. L. and D. T. Bell. 1976. Tree growth and mortality in the streamside forest. Castanea 41:34-41.

Jones, T. A. and S. C. Thomas. 2004. The time course of diameter increment to selection harvests in Acer saccharum. Can. J. Res./Rev. Can. Rech. For. 34:1525-1533.

Kimmins, J. P. 1987. Forest ecology. Macmillan Publishing Company. New York, NY, USA.

Kozlowski, T. T. 1997. Responses of woody plants to flooding and salinity. Tree Physiology Monograph 1:1-17.

Kozlowski, T. T. and S. G. Pallardy. 2002. Acclimation and adaptive responses of woody plants to environmental stresses. The Botanical Review 68:270-334.

Leopold, L. B., M. G. Wolman and J. E. Miller. 1964. Fluvial processes in geomorphology. W. H. Freeman, San Francisco, CA, USA.

Martens, D. M. 1993. Hydrologic inferences from tree-ring studies on the Hawkesbury River, Sydney, Australia. Geomorphology 8:147-164.

64 Megonigal, J.P., W.H. Conner, S. Kroeger and R.R. Sharitz. 1997. Aboveground production in southeastern floodplain forests: a test of the subsidy-stress hypothesis. Ecology 78:370-384.

Mitsch, W. J. and K. C. Ewel. 1979. Comparative biomass and growth of cypress in Florida wetlands. American Midland Naturalist 101:417-426.

Mitsch, W. J. and W. G. Rust. 1984. Tree growth responses to flooding in a bottomland forest in northeastern Illinois. Forest Science 30: 499-510.

Mitsch, W.J., J.R. Taylor and K.B. Benson. 1991. Estimating primary productivity of forested wetland communities in different hydrologic landscapes. Landscape Ecology 5:75-92.

Mitsch, W.J. and J.G. Gosselink. 2000. Wetlands, third edition. John Wiley & Sons, Inc., New York, NY, USA.

Mitsch, W. J. and L. Zhang. 2004. Wetland monitoring of the bottomland hardwood forest at the Olentangy River Wetland Research Park (Year 3 – 2003). p.137-147. In W.J. Mitsch, L. Zhang and C. Tuttle (eds.) Olentangy River Wetland Research Park at The Ohio State University, Annual Report 2003. Columbus, OH, USA.

Newbould, J. 1978. Methods for estimating the primary production of forests. Blackwell, Oxford, England.

Odum, E.P., J.T. Finn and E.H. Franz. 1979. Perturbation theory and the subsidy-stress gradient. Bioscience 29:344-352.

Phipps, R.L. 1979. Simulation of wetlands forest vegetation dynamics. Ecological Modelling 7:257-288.

Pollock, M. M., R. J. Naiman, and T. A. Hanley. 1998. Plant species richness in riparian wetlands- a test of biodiversity theory. Ecology 79:94-105.

Regent Instruments, Inc. 2002. WinDENDRO 2002 a,b. Regent Instruments, Inc., Quebec, Canada.

Robertson, A.I., P.Y. Bacon, and G. Heagney. 2001. The response of floodplain primary production to flood frequency and timing. Journal of Applied Ecology 38:126-136.

65 Teskey, R. O. and T. M. Hinckley. 1977a. Impact of water level changes on woody riparian and wetland communities, Vol. I: plant and soil response. U. S. Fish and Wildlife Service, Columbia, MO, USA. FWS/OBS-77/58.

Teskey, R. O. and T. M. Hinckley. 1977b. Impact of water level changes on woody riparian and wetland communities, Vol. III: the central forest region. U. S. Fish and Wildlife Service, Columbia, MO, USA. FWS/OBS-77/60.

Tockner, K., F. Malard and J.V. Ward. 2000. An extension of the flood pulse concept. Hydrological Process 14:2861-2883.

U.S. Forests Products Laboratory. 1974. Wood handbook: wood as an engineering material. USDA Agriculture Handbook No. 72. Washington, D.C., USA.

Whittaker, R.H. and G.M. Woodwell. 1968. Dimension and production relations of trees and shrubs in the Brookhaven Forest, New York. Journal of Ecology 57:155-174.

Zhang, L., W. J. Mitsch, V. Bouchard and K. Hossler. 2005. Sediment chemistry in a hydrologically restored bottomland hardwood forest in Midwestern U.S. Program and abstracts, restoration and design of ecosystems, fifth annual meeting, American Ecological Engineering Society, The Ohio State University, Columbus, OH, USA.

66 Cut #1 Cut #2 Cut #3

Clinton Park weir staff gauge station Levee

Existing Wetlands at the Olentangy River Wetland Research Park

Cut #4 LEGEND Elevation (m MSL) Area < 221.0 N 0.5 ha 221.0 - 221.3 0.6 ha 221.3 - 221.6 1.3 ha 221.6 - 221.9 1.4 ha > 221.9 1.4 ha Total 5.2 ha

Clinton Park weir gage elevations normal river pool 220.6 (m MSL) average river level 220.8 (m MSL)

0 50 100 meter

source: Dodson - Lindblom survey (1987) drawing by: N. Wang, ORWRP, OSU modified by C. Anderson , ORWRP, OSU (2005)

Figure 3.1. Map of the bottomland forest at the Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio, USA indicating site topography and levee breeches (Mitsch and Zhang 2004). Hydrologic restoration was conducted by breaching a levee (Cuts #1-3) along the north section and breaching the river bank at the south section (Cut #4).

Figure 3.2. Experimental layout at the ORWRP bottomland hardwood forest indicating the location and dimensions of tree plots and litter traps. Each tree plot was divided into four quadrants (NW, NE, SW, and SE) for placement of random litter traps including a fifth trap near the plot center.

80 Actual Normal 70

30 Precipitation (cm) Precipitation 20

10 69 0 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004

Figure 3.3. Quarterly-annual normal and recorded precipitation totals for Columbus, Ohio based on data collected from the Ohio Agriculture and Development Center weather station (www.oardc.ohio- state.edu/centernet/weather.htm). Precipitation totals reported for January-March, April-June, July- September and October-December of 1991-2004.

222.0

221.6

221.2

220.8 Olentangy River water level (m MSL) level (m Riverwater Olentangy 220.4 Jan-00 Jul-00 Jan-01 Jul-01 Jan-02 Jul-02 Jan-03 Jul-03 Jan-04 Jul-04 Jan-05

Figure 3.4. Hydrograph of river water levels (m above MSL) for the Olentangy River for 2001-2004 based on data collected at the Olentangy River Wetland Research Park (Mitsch and Zhang 2004).

1200 Litterfall Production Wood Production

) 1000 -1

yr 800 -2 600 400

ANPP (g m (g ANPP 200 0 1234 678910 5

North South Upl

Plot Number and Section

1200 Litterfall Production Wood Production 1000 ) 1 800 yr- -2 600

400 ANPP (g m (g ANPP 200

0 North South

Figure 3.5. Aboveground net primary productivity (ANPP), including litter-fall and wood production for a) tree plots in the north, south and upland sections, and b) mean (±1 SE) for north and south section plots for 2004-05. Error bars for the section means represent standard error for ANPP.

1100

) -1 1000 yr

-2 900

800

700 Total ANPP = 607 + 6*No. days 600 flooded R2=0.48, P=0.04 Total ANPP (g m 500 0 20406080

Total no. of days flooded (2003-2004)

Figure 3.6. Linear relationship between the number of days flooded (2003-2004) and aboveground net primary productivity for experimental plots in 2004.

a) )

-1 600 yr

-2 500 400 300 200 Wood prod. = 591 + 236 * Log (el. 100 var.) R2=0.73, P<0.01 0 Wood production (g m -2 -1.5 -1 -0.5 0 Log (elevation variance)

b) 1200 ) -1 1100 yr

-2 1000 900 800 700 Total ANPP = 1177 + 280 * Log (el. var.) 600 2

Total ANPP m (g Total R =0.63, P<0.01 500 -2 -1.5 -1 -0.5 0 Log (elevation variance)

Figure 3.7. Linear relationships between topographic variability (log-transformed elevation variance) and a) aboveground net primary productivity and b) wood production for experimental plots in 2004-05.

73 ) a)

-1 600 yr

-2 500 400 300 200 Wood prod. = 62 + 6.8 * Ttl. BA 100 R2=0.51, P=0.01 0 Wood production (g m 20.0 30.0 40.0 50.0 60.0 70.0 Total basal area (cm2 m-2)

b) 1100 ) -1 1000 yr -2 900 800 700 600 Total ANPP = 472 + 9.5 * Ttl. BA R2=0.68, P<0.01 Total ANPP m (g Total 500 20.0 30.0 40.0 50.0 60.0 70.0 Total basal area (cm2 m-2)

Figure 3.8. Linear relationships between total tree basal area and a) aboveground net primary productivity and b) wood production for experimental plot data in 2004-05.

7.00 North section (n=17) 6.00 South section (n=25) ) Upland section 5.00 (n=7)

4.00

3.00

2.00 Basal Area Increments (% Basal Increments Area 1.00 Pre-restoration Post-restoration 0.00

4 8 4 92 96 00 02 993 995 999 003 1991 19 1 199 1 19 1997 199 1 20 2001 20 2 200

Figure 3.9. Mean (±1 SE) BAI (%) for bottomland canopy trees from the north, south and upland sections from 1991 to 2004. The dashed line represents pre- and post-restoration periods.

a) Flooded sections 4.50 (n=19) 4.00 Upland section

) (n=2) 3.50 3.00

2.50 2.00 1.50 1.00 Basal Area Increments (% Increments Basal Area 0.50 Pre-restoration Post-restoration 0.00

1 3 4 5 8 9 9 9 9 9 02 9 9 9 0 1 1992 1 1 19 1996 1997 19 1999 2000 2001 2 2003 2004

b) 4.50 Flooded sections (n=6) 4.00 Upland section

) (n=2) 3.50 3.00 2.50

2.00 1.50 1.00 Basal Area Increments (% Increments Basal Area

0.50 Pre-restoration Post-restoration 0.00

1 2 3 7 9 1 9 9 9 9 9 0 9 9 9 0 19 1 19 1994 1995 1996 1 1998 1 2000 2 2002 2003 2004

Figure 3.10. Mean BAI(%) for a) boxelder (Acer negundo L.) and b) Ohio buckeye (Aesculus glabra Willd.) bottomland canopy trees in the flooded and upland sections from 1991 to 2004. The dashed line represents pre- and post-restoration periods.

a) 33.0

) 31.0 -1 yr 2 29.0

BAI (cm BAI 27.0 y = -0.08x2 + 1.47x + 24.08 R2 = 0.68; P=0.002 25.0 0 5 10 15 No. of days with river discharge >154 m3 sec-1 (over preceeding 2 years)

64.0 b) 62.0 ) -1 60.0 2yr 2 58.0 56.0

BAI (cm BAI 2 54.0 y = -0.13x + 2.81x + 47.65 R2 = 0.89; P=0.012 52.0 0 5 10 15 No. of days with river discharge >154 m3 sec-1 (over 2 year period)

Figure 3.11. Polynomial relationships between a) the number of days of river discharge >154 m3 sec-1 over the preceding two years and basal area increment (BAI) and b) the number of days of river discharge >154 m3 sec-1 and BAI over 2-yr periods from 1991-2004. Open symbols represent post-restoration years.

Plot environmental parameters Mean (±1 SE) Range

2004 No. of floods (total)* 4.4 ± 0.2 4 – 5 No. of floods (growing season)* 3.4 ± 0.2 3 – 4 Days connected with the river (total)* 21.1 ± 2.9 15 – 30 Days connected with the river (growing season)* 17.7 ± 2.7 7 – 26

2003 - 2004 No. of floods (total)* 11.1 ± 0.6 8 - 13 No. of floods (growing season)* 8.2 ± 0.6 6 - 10 Days connected with the river (total)* 37.8 ± 5.8 16 - 56 Days connected with the river (growing season)* 28.8 ± 4.9 11 - 44

2001 - 2004 No. of floods (total)* 16.2 ± 1.3 10 - 20 No. of floods (growing season)* 11.1 ± 0.9 8 -14 Days connected with the river (total)* 54.0 ± 8.1 22 - 81 Days connected with the river (growing season)* 38.3 ± 6.0 17 - 57

Mean plot elevation (m above MSL) 221.38 ± 0.07 221.08 – 221.86 Plot elevation variance 1.15 ± 0.43 0.21 – 4.67

Mean canopy cover (%) 81.7 ± 1.3 72.9 – 88.2 Total basal area (cm2 m-2) 39.0 ± 3.8 27.2 – 65.0

Note: Total year consists of 12 months (from preceding October-September). Growing season consists of 6 months (April-September) * Flood parameters do not include upland Plot #5 which was estimated to have flooded only once (in 2003) from 2001-2004.

Table 3.1. Synopsis of tree plot environmental variables used for regression analyses with forest productivity.

Importance Value Species (common name) North Sec. South Sec. Acer negundo L. (boxelder) 36.6 94.3 Acer saccharinum L. (silver maple) 15.0 8.9 Acer saccharum Marsh. (sugar maple) 7.1 9.2 Aesculus glabra Willd. (Ohio buckeye) 48.5 51.1 Asimina triloba (L.) Dunal (paw paw) 68.0 -- Celtis occidentalis Willd. (hackberry) 46.1 8.1 Fraxinus pennsylvanica Marsh. (green ash) 3.9 6.2 Gleditsia triacanthos L. (honey locust) -- 16.0 Juglans nigra L. (black walnut) 13.8 4.6 Lonicera maackii (Rupr.) Amur honeysuckle -- 7.0 Maclura pomifera (Raf.) (osage-orange) 3.8 -- Morus alba L. (white mulberry) 8.8 7.1 Morus rubra L. (red mulberry) 8.7 6.8 Platanus occidentalis L. (sycamore) 18.9 20.4 Populus deltiodes Bartr. Ex (cottonwood) 11.2 41.3 Prunus serotina Ehrh. (black cherry) -- 5.8 Salix nigra L. (black willow) -- 7.5 Ulmus americana L. (American elm) __9.5__ __6.0__ Total 300.0 300.0

Table 3.2. Importance value (= rel. density + rel. dominance + rel. frequency) of all tree species identified in the north and south sections of the ORWRP bottomland forest. Dominant species (Impt.value >35) are in bold.

Mean basal area increment Paired t-test Pre- Post- Section BAI restoration restoration (n=) parameter (1991-2000) (2000-2004) T-value P

North % 4.3 ±0.6 3.3 ±0.6 2.99 0.009 (n=17) cm2 yr-1 33.5 ±4.6 30.8 ±4.0 0.90 NS

South % 3.0 ±0.4 2.3 ±0.2 2.28 0.032 (n=25) cm2 yr-1 28.5 ±3.6 27.4 ±3.7 0.41 NS

Upland % 3.0 ±0.5 3.8 ±0.6 2.30 NS (n=7) cm2 yr-1 24.8 ±3.6 24.6.0 ±4.9 0.04 NS

Table 3.3. Results of paired t-tests for mean (±1 SE) basal area increment (BAI) (% and cm2 yr-1) of canopy trees pre- and post-restoration. NS denotes non-significant p-value.

Mean basal area increment Paired t-test Species BAI Pre-restoration Post-restoration (n=) parameter (1991-2000) (2000-2004) T-value P

A. negundo % 3.2 ±0.3 2.3 ±0.2 2.75 0.013 (n=19) cm2 yr-1 29.1 ±3.8 27.7 ±3.5 1.38 NS

A. glabra % 1.9 ±0.4 1.8 ±0.3 0.25 NS (n=6) cm2 yr-1 14.1 ±4.5 14.8 ±3.9 -0.35 NS

Table 3.4. Results of paired t-tests for mean (±1 SE) basal area increment (BAI) (% and cm2 yr-1) pre- and post-restoration for boxelder (Acer negundo L.) and Ohio buckeye (Aesculus glabra Willd.). NS denotes non-significant p-value.

CHAPTER 4

TEMPORAL AND SPATIAL DEVELOPMENT OF SURFACE SOIL CONDITIONS

AT TWO CREATED RIVERINE MARSHES

4.1 Abstract

The amount of time it takes for created wetlands to develop soils comparable to natural wetlands is relatively unknown. Surface soil changes over time were evaluated in

two created wetlands (~1 ha each) at the Olentangy River Wetland Research Park in

Columbus, Ohio. The two wetlands were constructed in 1993 to be identical in size and

geomorphology, and maintained to have the same hydrology. The only initial difference

between the wetlands was that one was planted with native macrophytes while the other

was not. In May 2004, soil samples were collected (10 years and 2 months after the

wetlands were flooded) and compared to samples collected in 1993 (after the wetlands

were excavated but prior to flooding) and 1995 (18 months after the wetlands were

flooded). In all three years, soils were split into surface (0-8 cm) and subsurface (8-16

cm) depths and analyzed for soil organic matter, total C, total P, available P,

exchangeable cations and pH. Soils in the two wetlands have changed substantially

82 through sedimentation and organic accretion. Between 1993 and 1995, soils were most

influenced by the deposition of senescent macroalgae, the mobilization of soluble

nutrients and the precipitation of CaCO3. Between 1995 and 2004, soil parameters were influenced more by the deposition of organic matter from colonized macrophyte communities. Mean percent organic matter at the surface increased from 5.3 ±0.1% in

1993, 6.1 ±0.2% in 1995, to 9.5 ±0.2% in 2004. Mean total P increased from 493 ±18 μg

g-1 in 1993, 600 ±23 μg g-1 in 1995, to 724 ±20 μg g-1 in 2004. Spatial analyses of

percent organic matter (a commonly used indicator of hydric soil condition) at both

wetlands in 1993, 1995 and 2004 showed that soil conditions have become increasingly

more variable. High spatial structure (autocorrelation) between data points was detected

in 1993 and 2004, with data in 2004 exhibiting a much higher overall variance and

narrower range of spatial structure than in 1993.

4.2 Introduction

Wetlands are constructed throughout the United States to provide landscape functions

such as wildlife habitat, flood attenuation, and water quality enhancement (Mitsch and

Gosselink 2000). Where regulatory requirements stipulate monitoring of wetland

creation areas, hydrology and vegetation are usually used as indicators of wetland

condition. Soils are often the least considered component of wetland systems despite

their importance in providing the substrate for many of the biological and chemical

processes that make them valuable components to the landscape (Vepraskas and Faulkner

2001, Collins and Kuehl 2001). There have been an increasing number of studies

conducted to evaluate soil conditions in created wetlands. Many of the studies have been

83 designed to compare the soils of created wetlands to natural reference wetlands (Bishel-

Machung et al. 1996, Shaffer and Ernst 1999, Zedler and Callaway 1999; Nair et al. 2001,

Campbell et al. 2002) with most finding some progressions toward natural wetland soil conditions but with substantial differences in many key characteristics (e.g., lower soil organic matter concentrations, coarser texture, and dissimilar nutrient concentrations and pH).

When terrestrial soils become flooded, there are several biogeochemical

transformations that can occur over different time intervals. After only a few days of

flooding, oxygen in the soil column becomes depleted and microbial activity will be dominated by facultative and strict anaerobes (Mitsch and Gosselink 2000). Soil colors will become darker as reduced Fe and Mn are transported out of the soil column during flooded conditions. Soils with chroma values < 2 are used to indicate the presence of

hydric soil conditions (Tiner 1999). Flooding also influences soil P availability due to its

release into the water column (Sanyal and De Datta 1991). Longer-term changes in soil condition are influenced by the buildup of soil organic matter at the surface caused by the reduced rate of decomposition. The accumulation of soil organic matter has been identified as an indication of soil maturity in created wetlands because of the time required for it to develop (Craft 2001, Nair et al. 2001). Several important

biogeochemical processes associated with wetlands (e.g. denitrification) are dependent

upon adequate soil carbon being present (Mitsch and Gosselink 2000).

While several studies have evaluated temporal changes in the soil organic matter of

created wetlands (Bishel-Machung et al. 1996, Nair et al. 2001, Anderson and Cowell

2004), few have evaluated the spatial patterns that occur over time. This is partially

84 because of the explicit sampling design that is required to evaluate spatial dynamics.

Spatial patterns associated with natural wetland soil characteristics such as P enrichment

in the Everglades (DeBusk et al. 2001) and P-sorption capabilities in North Carolina

floodplains (Bruland and Richardson 2004) have been studied. Changes in how soil

properties were distributed within a created wetland were observed after two years in

response to flooding at the Des Plaines River wetlands near Chicago, Illinois (Fennessy

and Mitsch 2001). They found that spatial variability of soil organic C and exchangeable

nutrient concentrations as measured by the range of autocorrelation influence declined

after two years of flooding.

At the Olentangy River Wetland Research Park (ORWRP) in Columbus, Ohio, two 1-

ha experimental wetlands were constructed in 1993. One was planted with native

macrophytes and the other was not. Extensive soil surveys were conducted at the two

wetlands in August 1993 (after excavation but prior to flooding), September 1995 (18

months after flooding) and May 2004 (10 years and 2 months after flooding) to evaluate

changes in response to flooding. Over the last ten years, several investigations have identified the rapid development of a sediment layer, and significant increases in soil organic C, Ca, Fe, P and total C (Nairn 1996, Liptak 2000, Harter and Mitsch 2003).

Initial accumulations were attributed to the autochthonous production of dense algae mats

(Wu and Mitsch 1998) and allochthonous import of sediment (Harter and Mitsch 2003).

After the third year, both wetlands had developed significant cover by macrophyte

communities, which are now considered the primary contributor to soil organic matter.

This study represents the first to examine changes in soil condition at the two wetlands

since its creation along with changes in spatial variability. Detailed descriptions of the

85 hydrologic, biogeochemical, and ecological patterns of these experimental wetlands are

given by Metzker and Mitsch (1997), Mitsch et al. (1998, 2005a,b,c), Kang et al. (1998),

Koreny et al. (1999), Nairn and Mitsch (2000), Spieles and Mitsch (2002a,b, 2003), Ahn

and Mitsch (2002), Anderson et al. (2002), Selbo and Snow (2004), and Zhang and

Mitsch (2005).

The first objective of this study was to compare soil data collected in 1993, 1995 and

2004 to evaluate changes at the soil surface that have occurred as a result of the created

riverine-wetland conditions. Given the high productivity and flooded conditions, we

hypothesized that the wetland soil surface has substantially increased in its concentration

of organic matter and nutrients associated with organic matter (organic C, N, P and

exchangeable cations). Our second objective was to compare the spatial patterns of soil

organic matter concentrations in samples collected in 1993, 1995 and 2004. Starting with

the antecedent soil conditions (1993), we expected to see an increase in the concentration

and spatial variability of organic matter.

4.3 Methods and materials

4.3.1 Study area

The study was conducted at the Olentangy River Wetland Research Park (ORWRP)

on The Ohio State University campus in Columbus, Ohio, USA (latitude N40.021◦, longitude E83.017◦). The ORWRP is a 10-ha facility located along the Olentangy River

and was constructed on abandoned agricultural land. Underlying soils in this area are

alluvial floodplain soils, comprised of the Ross and Eldean series (classified as a Cumlic

Hapludoll), which include silt loams, silt clay, and clay loams (Mcloda and Parkinson

86 1980). Two 1-ha experimental marshes were excavated at the ORWRP in 1993 and

flooded in March 1994 with pumped Olentangy River water. The two wetlands were

built and managed identically with Olentangy River water being pumped at a similar rate

(typically ~25 m yr-1) throughout their ten-year history. As part of a long-term study, the

only difference between wetlands was that the western marsh (Wetland 1) was planted

with native, wetland vegetation while the eastern marsh (Wetland 2) was left unplanted

(Mitsch et al. 2004) (Fig. 4.1). Based on their topography, both wetlands have developed

two distinct cover zones: a shallow, emergent vegetation (EM) zone and three deeper,

open-water (OW) sub-basins spaced longitudinally along each wetland (Fig. 4.2). The

EM zones were constructed approximately 0.3 m below natural grade and the OW basins

were typically 0.6 m below grade. In the first three years, both wetlands were similar in

form with large areas of open water gradually colonizing with macrophyte cover in the

EM zones, predominantly Schoenoplectus tabernaemontani (C.C. Gmel) Palla. However,

between the years of 1998 and 2001, Wetland 2 became dominated by dense stands of

Typha spp. (mostly Typha angustifolia L., Selbo and Snow 2004) while Wetland 1 maintained a more mixed community assemblage. Because of their depth, the OW zones have only supported sparse amounts of emergent macrophytes, but have supported macroalgae and other aquatic vegetation (e.g., Ceratophyllum sp. and Lemna sp.,

Anderson and Mitsch 2003). Although both wetlands were excavated to be 1-ha in size, after ten years of peripheral shrub encroachment, the combined marsh area of Wetland 1 and 2 in 2004 was approximately 0.81 and 0.88 ha, respectively with the OW zone covering approximately 29 and 28% of each wetland, respectively.

4.3.2 Soil sampling design

Soil sampling in 1993, 1995 and 2004 was conducted based on a 10-m grid system

marked at each intersection point with a permanently installed 2-cm diameter PVC pole

(Fig. 4.2). In 1993 (after wetland construction, but before flooding) and 1995 (18 months

after flooding), soil samples were collected at the same 43 intersection points (Fig. 4.2) at

a depth of 0-8 cm and 8-16 cm. In 2004, a total of 127 intersection points were used to collect samples at 0-8 cm and 8-16 cm depths (Fig. 4.2).

Sampling methods described below are specific to the 2004 sampling period, but were designed to be consistent with methods used in 1993 and 1995 (Nairn 1996). Water depths were lowered to minimize standing water at each grid point and facilitate soil extraction. At each grid point, soils were collected 0.5 m east of the field marker. Soils were collected using a 10-cm diameter steel soil-corer, carefully removed, and split into

0-8 and 8-16 cm sections using a sharp knife. The 0-8 cm section was then halved length wise and stored in separate water-tight freezer bags. Because most 8-16 cm sections were

typically dense clay, this section was split into quarters and two of the quarter-sections

were placed in separate plastic freezer bags. Soil remnants were replaced into the sample

hole. For each sample, the hue, value and chroma were determined using a Munsell

Color Chart and other visual characteristics were noted. Because of the dense consistency

of the antecedent soil surface, the development and boundary of the accreted sediment-

layer was usually apparent. When it was, the depth was measured to the nearest 0.5 cm.

Each sample section was placed in a plastic freezer bag and kept in an ice-packed cooler

88 until being returned to the laboratory where they were refrigerated at 4 ûC until laboratory analysis.

4.3.3 Physical and chemical soil analyses

One section of each soil sample was weighed and placed in a drying-oven at 105 ûC

for five days or until constant mass occurred. Soil sections were reweighed to determine

soil moisture content and bulk density. The second section of each soil sample was kept

in its field-moist, natural condition and completely homogenized by hand. A 30-g

subsample of each sample was air-dried at room temperature, ground using a pestle and

mortar, and passed through a 2-mm sieve. Duplicate subsamples (approximately 10 g

each) were placed in a crucible, oven-dried at 60 ûC overnight, weighed, and ignited in a

muffle furnace at 550ûC for 1 hour. The post combustion material was reweighed and the

duplicates averaged to estimate the percent organic matter of the soil.

A second subsample from the field-moist section was used to characterize soils for

various chemical parameters. For each year, samples (at 0-8 and 8-16 cm depths) were

collected and analyzed from the same grid points. Samples were selected to analyze

chemical conditions over an even spatial distribution and to be proportionate among the

cover zones. A total of 168 samples [56 per year based on 2 samples (0-8 and 8-16 cm)

collected at 28 grid points] were analyzed for available P by the Bray-P1 extraction (Kuo

1996), exchangeable K, Ca, and Mg by 1M ammonium acetate extraction (Warncke and

Brown 1998), and pH (Thomas 1996). A total of 108 of these samples [36 per year based

on 2 samples (0-8 and 8-16 cm) collected at 18 grid points] were further analyzed for

total C by combustion (ISO, 1995; AOAC, 1989) and total P by digestion with

89 HClO4/HNO3 followed by inductively coupled plasma emission spectrometry (Sommers and Nelson 1972).

4.3.4 Temporal and geostatistical statistical analyses

Because several of the soil parameters had unequal variances and could not be transformed to fit a normal distribution, mean comparison of each soil parameter in 1993,

1995 and 2004 was conducted using nonparametric Friedman Two-Way Analysis of

Variance of repeated measure with post hoc comparison of years conducted using

Wilcoxon Signed Ranks Test. The Friedman test was used to strictly evaluate changes over time at each repeatedly sampled grid point (no intra-annual comparisons were considered) therefore the potential ramifications of using non-independent data were minimized. All tests were conducted using Systat v.10.2 (Systat Software Co. 2002). For each statistical test, differences were considered significant at p<0.05 and highly significant at p<0.01 with a Bonferroni adjustment for individual Wilcoxon tests.

Changes in the spatial pattern of soil organic matter were evaluated for both wetlands using data collected in 1993, 1995 and 2004. GS+ Software (Version 7.0) (Gamma

Design Software 2004) was used to assess for autocorrelation based on the semivariance of paired groups of data points within each wetland (each wetland was analyzed separately). Isotropic variograms were used to detect semivariance and were constructed using 20m interval classes over 100m distances for the 1993 and 1995 data, and 10m interval classes over 70m distances for the 2004 data. H-scatterplots were used to detect for outliers or aberrant data that may have had excessive influence on the model parameters (Isaaks and Srivastava 1989). Variograms consist of a graphical output in

90 which the semivariance is measured at increasingly further interval distances. When autocorrelation occurs, the level of variance between interval classes increases and eventually reaches an asymptote and levels off, representing the extent of autocorrelation

(Isaaks and Srivastava 1989). Characteristics of the variogram graph include 1) the

nugget variance (C0) which is the experimental variance unaccounted for by the spatial

model, 2) the sill (C0 + C) which is the total variance as measured at the asymptote of the

variogram, and 3) the range (A0) which is the spatial distance in which autocorrelation is

detected.

The spatial structure detected through the variogram was used to conduct kriging

analyses, which is an unbiased procedure that uses the modeled spatial relationship to

interpolate values between data points (Isaaks and Srivastava 1989). The interpolated

data from each kriging analysis were used to calculate frequency distributions of soil

organic matter in Wetland 1 and 2 for 1993, 1995 and 2004. The interpolated data were

also used with the GS+ software to illustrate kriging maps for comparisons between

wetlands and years.

4.4 Results

4.4.1 Temporal changes to soil properties

Since wetland construction in 1993, soil conditions at the surface of both wetlands

have developed substantially through sedimentation and organic accretion. Soils in both cover zones generally consisted of an unconsolidated sediment layer atop of the much denser, clay layer (the antecedent soil surface). Mean depth of the sediment layer was

9.3 ±0.4 cm and ranged between 1.5 and 22.0 cm throughout both wetlands (Appendix C,

91 Table C.1). As a result, conditions at the 0-8 cm depth showed the most changes over time (Fig. 4.3). In the open water (OW) zones, the sediment layer tended to be deeper

(ranging between 8.0 and 22.0 cm), was grey-black in color, and very homogeneous with

very fine particulate matter. The consistency of the layer was almost gelatinous in nature

suggesting the formation of a gyttja layer (Wetzel, 2001). The sediment layer in the

emergent (EM) zones was more cohesive and heterogeneous with samples containing

variable amounts of fine mineral/organic sediment, undistinguishable macrophyte detritial

matter, living macrophyte roots/rhizomes and soil fauna. Short term buildup of the

sediment layer (between 1993 and 1995) occurred rapidly and while not measured

systematically, a range of 0-15 cm was estimated by 1995 (Nairn 1996). Conditions in

the antecedent soil layer consisted were much denser with samples in the EM zones

occasionally containing fine root material and oxidized rhizospheres. All soil samples (0-

8 and 8-16 cm depths) analyzed in 2004 had a chroma value <2, compared to 51% in

1995 (78% at the 0-8 cm depth and 24% at the 8-16 cm depth), and none in 1993.

Since 1993, the mean percent organic matter at the 0-8 cm depth has increased from

5.3 ±0.1% (pre-wetland), to 6.1 ±0.1% in 1995 (18 months after creation), to 9.5 ±0.5% in 2004 (10 years after creation) (Fig. 4.3a; Appendix C, Table C.1,2). Similarly, total C increased from 1.55 ±0.05% in 1993, to 2.03 ±0.10% in 1995, and to 3.70 ±0.19% in

2004 (Fig. 4.3b). Total P also increased over time in the 0-8 cm depth (Fig. 4.3c). The concentration of available P declined significantly in 2004 (Appendix C, Table C.3) at both 0-8 and 8-16 cm depths after significant increases were observed between 1993 and

1995 (Fig. 4.3d). Exchangeable Ca had a significant increase detected at 0-8 cm depths in 1995 and then decreased in 2004, but was still significantly higher than concentrations

92 in 1993 (Fig. 4.3e). Exchangeable K and Mg concentrations have risen significantly at the 0-8 cm depth since 1995 (Fig. 4.3f,g) and soil pH has continually increased since

1993 at both depths (Fig. 4.3h).

Subsurface soil conditions (8-16 cm depth) had fewer significant changes between years (Fig. 4.3; Appendix C; Table C.1-3). No substantial changes between years were detected for percent organic matter (Fig. 4.3a), exchangeable Ca (Fig. 4.3e) and exchangeable Mg (Fig. 4.3g). Percent total C (Fig 4.3b) and exchangeable K (Fig 4.3f) increased since 1993 but only significantly after ten years. Other soil attributes such as available P (Fig. 4.3d) and soil pH (Fig. 3h) showed temporal changes that were similar to those observed at the 0-8 cm depth.

4.4.2 Spatial characteristics and changes of soil organic matter

Variogram characteristics using soil organic matter data from 1993, 1995 and 2004 were evaluated for spatial structure (Table 4.1). For the 2004 data, two outliers from

Wetland 1 and two from Wetland 2 were identified through a review of an h-scatterplot and removed because of their excessive influence on the model parameters. Strong autocorrelation was detected for Wetland 1 and 2 in 1993 and 2004 with no spatial structure detected in 1995. Both wetlands had similar changes between 1993 and 2004, with an overall increase in variance (represented by the sill value, C0 + C) and substantial decreases in the range at which autocorrelation was detected (A0) (Table 4.1). The range of autocorrelation detected in Wetland 1 and 2 decreased by 90 and 60%, respectively

(Table 4.1). The proportion of variance explained by autocorrelation was moderate to

93 high in both wetlands, increasing from 0.59 to 0.91 in Wetland 1 and decreasing from

0.99 to 0.57 in Wetland 2.

Because no spatial structure was detected for either wetland in 1995, a kriging

analysis could not be conducted for the soil organic matter concentration data for that

year. In its place, we used an inverse distance weighing method (using the GS+

software) which interpolates on the basis that grid points closer together will be more

related than those farther apart (Isaaks and Srivastava 1989). This method estimates unsampled grid point values by weighing each sampled grid point so they are inversely proportional to the distance of the point being estimated.

Review of the frequency distributions (Fig. 4.4) and maps (Fig. 4.5) of spatially

interpolated data showed that soil organic matter at the 0-8 cm depth exhibited a fairly

even distribution in 1993 and 1995 compared to conditions in 2004. Soil organic matter in 1993 and 1995 were similar between wetlands, with Wetland 1 having a slightly broader range in 1993 than Wetland 2 (Fig. 4.4). Both wetlands showed an increase in the percent organic matter over the ten year period and developed concentration ranges that appeared comparable. Evaluating the 2004 field data, soil organic matter ranged from 4.2 to 15.5% in Wetland 1 and 4.5 to 19.1% in Wetland 2. Kriging analysis tends to suppress the range of interpolated values; therefore extreme high and low measurements tend to be less represented by this procedure. Nevertheless, it was apparent that a wide range of soil organic matter conditions existed for both wetlands in 2004. Based on 0.5 percent organic matter intervals, the frequency distribution did not exceed 20% of the total at any point for either wetland (Fig. 4.4). Wetland 2 showed a distribution that was slightly higher with a narrower range compared to Wetland 1. Evaluating the kriging

94 maps (Fig. 4.5), it is apparent that the amount and distribution of soil organic matter has

increased with time. In 2004, both wetlands tended to have the greatest concentrations of

organic matter along the wetland periphery (Fig. 4.5c) in areas associated with the EM

zones. Conditions in Wetland 1 were patchier and exhibited a broader range than

Wetland 2. It should be mentioned that part of the differences between years may be

attributed to the higher intensity of sampling conducted in 2004 (n=127) compared to

1993 (n=43) and 1995 (n=43). However, given the relatively homogeneous conditions

observed in 1993 and 1995, it is unlikely that more intensive sampling during those years

would have revealed substantial differences in the kriging maps or frequency distributions. The wetlands also showed differences in organic matter concentrations

from inflow to outflow. Evaluating the Wetland 2 kriging map, its greatest levels of soil

organic matter tended to be in the northern half of the wetland (closer to the inflow) while

this trend was less apparent in Wetland 1.

4.5 Discussion

4.5.1 Temporal changes to soil properties

Several changes in soil condition observed in 1993, 1995 and 2004 were caused by

changes in the source of contributing biomass to soil organic matter. Short term soil

changes (between 1993 and 1995) were influenced by the rapid growth and senescence of

macroalgae during the first season (1994) (Wu and Mitsch 1998). However after 1995,

the widespread colonization of macrophytes (Mitsch and Zhang 2004) became the

prevalent source of soil organic matter. Percent total C followed the same pattern as

percent organic matter, however, in 2004 there was a significant increase at the 8-16 cm

95 depth. This increase was due to extremely high concentrations of inorganic C detected in

the sediment layer of the OW zones in both wetlands. It has been determined that a

substantial accumulation of CaCO3 has occurred in these portions of the wetlands as a result of high algal photosynthesis which effectively alters water column pH and elicits

the precipitation of CaCO3 (Liptak 2000). Mean inorganic C soil concentrations of 1.0

and 1.4% have been observed in the OW zones of Wetland 1 and 2, respectively while

macrophyte colonization has effectively precluded this process in the EM zones

(unpublished data).

Available P concentrations in the created wetland soils were influenced by the release

of P in response to flooded conditions. The highest concentrations of available P were observed at the surface and subsurface soils in 1995 and likely reflected the initial soil response to flooded conditions. It has been shown that flooding previously terrestrial soils can cause created wetlands to be a source rather than a sink for P (Newman and

Pietro 2001). Upon submergence, P can be released through the reduction of Fe(III), and after subsequent pH increases, further desorption of P from clays, Al oxides, and Fe oxide surfaces may also occur (Sanyal and De Datta 1991). In the case of the ORWRP wetlands, available P actually increased 11-16% during the first 18 months after flooding

(Nairn 1996). Since 1995, available P concentrations declined at both surface and subsurface depths and were substantially less than pre-wetland (1993) levels.

In contrast to available P, total P concentrations at the soil surface increased continually. Total P accumulation in created wetlands has been shown to be influenced by a combination of sorption, organic accumulation, sedimentation and precipitation

(Craft 1997). At the ORWRP wetlands, high algal photosynthesis has elicited significant

96 co-precipitation of CaCO3 and P, particularly in the open water zones where algal

productivity is the greatest (Liptak 2000). This was especially the case between 1993 and

1995 when high algal productivity occurred throughout both wetlands. During this 18

month period, the average concentration of total P increased approximately 100 μg g-1 (or

60.2 μg g-1 yr-1); however since 1995 the concentration of total P has only increased by

another 120 μg g-1 (or 14.1 μg g-1 yr-1). This can be attributed to the colonization of

macrophytes in the EM zone and the overall suppression of algal productivity over most

of the wetland area. Also, sedimentation rates have reduced after ten years of flooding

compared to the initial years of wetland formation (Nairn 1996, Harter and Mitsch 2003).

Temporal changes in exchangeable cation concentrations were likely influenced by

several factors over different time scales, including short term mobilization in response to flooding (1993-1995), sedimentation and sorption (1993-2004), and the higher exchange capacity associated with organic accumulation (1993-2004). Concentration of

exchangeable Ca at the soil surface was greatest in 1995 and reflects the high deposition

of CaCO3 that occurred during this time period (Liptak 2000). The trend in concentration

between years is similar to that exhibited by available P which is not surprising because

of the shared biogeochemical processes that influenced both of them. Between 1993 and

2004, the steady increase in concentration of exchangeable K in both surface and subsurface soils was likely influenced by increased soil organic matter accumulation.

Exchangeable Mg showed significant but relatively small changes between years which is indicative of its long residence time and conservative nature in response to solubility equilibria (Hem 1989). Soils also exhibited a continual significant increase in pH between 1993 and 2004. A convergence of soil pH to neutral is the typical response that

97 mineral soils have when they are flooded (Ponnamperuma 1972) and this appears to be

happening at the ORWRP.

4.5.2 Spatial characteristics and changes of soil organic matter

The spatial changes seen between 1993, 1995 and 2004 illustrated that soil conditions

have become increasingly variable. There are several factors related to wetland

morphology that likely contributed to the strong spatial structure detected in 2004

(Johnston et al. 2001). Location within the two cover zones would have had a much

greater influence on soil development leading up to the 2004 sampling compared to

conditions in 1993 and 1995. Longer-term differences in water level and the frequency of

inundation can influence macrophyte colonization (Grace and Wetzel 1981) and

productivity (Newman et al. 1998). Standing water will also influence soil conditions

such as oxygen availability. Based on the 2004 soil organic matter maps (Fig. 4.5), it was

apparent that the greatest concentrations of organic matter were detected along the

wetland periphery (EM zones) with lesser concentrations in the central portions of the

wetland (OW zones), even though the OW zones have remained inundated longer since

1993. Other investigators have also found that created wetland soils supporting emergent

vegetation accumulated more organic matter than deep areas that were devoid of

vegetation (Shaffer and Ernst 1999).

Sediment accumulation and depth played a key role in organic matter concentrations

at both wetlands. Net sediment accumulation has been similar between wetlands with the

greatest accumulation occurring in the OW zones (unpublished data). However, as

indicated in the results of this study, the concentration of organic matter was greatest in

98 the EM zones suggesting that these areas are heavily influenced by the annual deposition of autochthonous organic matter. A review of annual vegetation community maps

(Mitsch and Zhang 2004) has shown that some of the areas with the greatest concentrations of organic matter in both wetlands have been dominated by Typha spp.

(the most productive macrophyte community) over several years. It is uncertain how much movement of organic matter occurs from the EM to the OW zones, but based on the refined quality of the organic matter in the OW zones, much of it appears to be allochthonous material or decomposed algal material, and therefore organic matter accumulation within the two cover zones may be two separate processes.

The 2004 organic matter concentration maps (Fig. 4.5) also illustrated a trend of decreasing soil organic matter from inflow to outflow in Wetland 2 that may have contributed to the overall spatial structure. Larger macrophyte production has been detected in the northern half of the wetland during previous years (Mitsch et al. 2004) and it is hypothesized that this was elicited by greater nutrient availability closer to the wetland inflow than the outflow. However, there are other circumstances associated with

Wetland 2 that may also explain this condition. In 2003, most of the southern section of

Wetland 2 was denuded of vegetation by apparent muskrat (Ondatra zibethicus) activity

(Mitsch and Zhang 2004). Along with the physical upheaval of the soil, large sections of

Wetland 2 were left without any recruitment of detritial matter in the year leading up to the 2004 soil sampling and this may have reduced soil organic matter concentrations in this area compared to the northern sections.

On a smaller scale, specific topographic conditions also influenced the reported soil organic matter concentrations. In both wetlands, areas sampled near the crest of the OW

99 sub-basins tended to have the least amount of sediment accretion and consequently the

lowest organic matter concentrations. It appeared that the unconsolidated sediment that

tends to accumulate in the OW zones may be susceptible to sloughing down into the

deeper portions of the sub-basin or being transported elsewhere during high flow. All of these factors contributed to the patchy spatial structure in the two experimental wetlands.

Spatial trends in the antecedent soil organic matter (represented by the 1993 data)

showed moderate to strong autocorrelation that had a much broader range of influence compared to 2004 conditions. This was expected from exposed subsurface soils that were developed by broad influencing factors such as climate and geology, rather than the new surface soil conditions incurred at the two 10-yr old wetlands. The lack of a detected spatial structure in 1995 may be attributed to the soil sampling design, specifically the distance between sampling points (typically 20 m apart) which may have been too far apart to detect spatial structure. However conditions at the wetlands during this time may also have contributed to an overall lack of structure. After flooding the wetlands in 1993, the productivity and deposition of autochthonous metaphyton was identified as the major source of organic matter in the newly forming soil surface (Nairn 1996, Wu and Mitsch

1998). Aerial photography from summer 1994 (Mitsch and Zhang 2004) indicated that algae coverage throughout both wetlands was extensive and may have provided a uniform influence on surface organic matter concentrations. Because of this, it is reasonable to expect that spatial structure was minimal after the initial two years of flooding. After two years of flooding at the Des Plaines wetlands in Illinois, Fennessy and Mitsch (2001) found that the range of autocorrelation decreased for most soil attributes. The most substantial changes were seen in exchangeable P and K, however a modest decrease in

100 organic C was also detected (196m in 1988 to 175m in 1990). They also found that after

two years, the combined variance (nugget and sill) had decreased in most soil attributes

after flooding. However after ten years at the ORWRP wetlands, we found that variance had increased substantially for surface soil organic matter concentrations.

4.6 Conclusions

Ten-year changes at the soil surface of the ORWRP experimental wetlands have been extensive because of sedimentation, organic accretion and the precipitation of CaCO3.

Changes were detected primarily in the surface soils which are most representative of the sediment layer. Subsurface soils, that often included much of the antecedent soil layer, changed less between 1993 and 2004. Soil organic matter concentrations were greatest in the emergent vegetation zones and showed considerable variability throughout both wetlands. As a result of ten years of wetland conditions, the spatial structure of soil organic matter in the wetlands has changed dramatically. Before the wetlands were flooded, soil conditions showed spatial structure with relatively low variance and a high range of autocorrelation. Ten years later, soil conditions have become much more variable. Soil data from 2004 exhibited strong autocorrelation with much shorter range of autocorrelation. The occurrence of macrophytes and the variable distribution of sediment were likely the primary reasons for the patchier conditions.

4.7 Acknowledgements

Funding for this project was provided through an Ohio Agricultural Research &

Development Center Graduate Research Enhancement Grant, the USDA (Grant No.

101 2002-35102-13518) and from support by the School of Natural Resources at The Ohio

State University. Statistical advice was provided by the Ohio State University, Statistics

Consulting Service. Warren Dick and two anonymous reviewers provided comments that

greatly improved this chapter. Olentangy River Wetland Research Park reprint number

05-008.

4.8 Literature cited

Ahn, C. and W.J. Mitsch. 2002. Scaling considerations of mesocosm wetlands in simulating large created freshwater marshes. Ecological Engineering 18: 327-342.

Anderson, C.J. and B.C. Cowell. 2004. Mulching effects on the seasonally flooded zone of west-central Florida, USA wetlands. Wetlands 24:811-819.

Anderson, C.J. and W.J. Mitsch. 2003. Open-water autotrophs: biomass and distribution in deepwater basins of two experimental wetlands. p. 41-44. In: W.J. Mitsch, L. Zhang and C.J. Anderson (eds.) Olentangy River wetland Research Park at The Ohio State University Annual Report 2002. The Ohio State University, Columbus, OH.

Anderson, K.L., K.A. Wheeler, J.A. Robinson, and O.H. Tuovinen. 2002. Atrazine mineralizaation potential in two wetlands. Water Research 36: 4785-4794.

AOAC Official Methods of Analysis. 1989. Method 990.03. Protein(crude) in Animal Feed Combustion Method(Dumas method). 17th Edition 2002. Reference: JAOAC 72, 770.

Bishel-Machung, L., R.P. Brooks, S.S. Yates, and K.L. Hoover. 1996. Soil properties of reference wetlands and wetland creation projects in Pennsylvania. Wetlands 16:532- 541.

Bruland, G.L. and C.J. Richardson. 2004. A spatially explicit investigation of phosphorus sorption and related soil properties in two riparian wetlands. J. Environ. Qual. 33:785-794.

Campbell, D.A., C.A. Cole, and R.P. Brooks. 2002. A comparison of created and natural wetlands in Pennsylvania, USA. Wetlands Ecology and Management 10:41-49.

102 Collins, M.E. and R.J. Kuehl. 2001. Organic matter accumulation and organic soils. In: J.L. Richardson and M.J. Vepraskas (eds.) Wetland Soils: Genesis, Hydrology, Landscapes and Classification. Lewis Publishers, Boca Raton, FL.

Craft, C.B. 1997. Dynamics of nitrogen and phosphorus retention during wetland ecosystem succession. Wetlands Ecology and Management 4:177-187.

Craft, C.B. 2001. Biology of wetland soils. In: J.L. Richardson and M.J. Vepraskas (eds.) Wetland Soils: Genesis, Hydrology, Landscapes and Classification. Lewis Publishers, Boca Raton, FL.

DeBusk, W.F., S. Newman, and K.R. Reddy. 2001. Spatio-temporal patterns of soil phosphorus enrichment in Everglades Water Conservation Area 2A. J. Environ. Qual. 30:1438-1446.

Fennessy, M.S. and W.J. Mitsch. 2001. Effects of hydrology and spatial patterns of soil development in created riparian wetlands. Wetlands Ecology and Management 9:103-120.

Gamma Design Software. 2004. GS+ User’s Guide Version 7: GeoStatistics for the Environmental Sciences. Gamma Design Software, LLC. Plainwell, MI.

Grace, J.B. and R.G. Wetzel. 1981. Habitat partitioning and competitive displacement in cattails (Typha): experimental field studies. The American Naturalist 118:463-474.

Harter, S.K. and W.J. Mitsch. 2003. Patterns of short-term sedimentation in a freshwater created marsh. J. Environ. Qual. 32:325-334.

Hem, J.D. 1989. Study and Interpretation of the Chemical Characteristics of Natural Water, Third Edition. U.S. Geological Survey Water- Supply Paper 2254. Washington, D.C.

International Standard, ISO 10694:1995(E). Soil quality – Determination of organic and total carbon after dry combustion (elementary analysis). International Organization for Standardization. Geneva, Switzerland.

Isaaks, E.H. and R.M. Srivastava. 1989. An Introduction to Applied Geostatistics. Oxford University Press, Inc. New York, NY.

Johnston, C. A., S. D. Bridgham, and J. P. Schubauer-Berigan. 2001. Nutrient dynamics in relation to geomorphology of riverine wetlands. Soil Sci. Soc. Am. J. 65:557-577.

Kang, H., C. Freeman, D. Lee, and W.J. Mitsch. 1998. Enzyme activities in constructed wetlands: Implication for water quality amelioration. Hydrobiologia 368:231-235.

103 Koreny, J.S., W.J. Mitsch, E.S. Bair, and X. Wu. 1999. Regional and local hydrology of a created riparian wetland system. Wetlands 19:182-193.

Kuo, S. 1996. Phosphorus. p. 894-895. In: Methods of Soil Analysis, Part 3—Chemical Methods. Soil Science Society of America, Madison, WI.

Liptak, M. 2000. Water column productivity, calcite precipitation, and phosphorus dynamics in freshwater marshes. PhD. Dissertation. The Ohio State University, Columbus, OH.

Mcloda, N. A. and R. J. Parkinson. 1980. Soil Survey of Franklin County, Ohio. USDA- SCS. US Government Printing Office, Washington, D.C., USA.

Metzker, K. and W.J. Mitsch. 1997. Modelling self-design of the aquatic community in a newly created freshwater wetland. Ecological Modelling 100:61-86.

Mitsch, W.J., X. Wu, R.W. Nairn, P.E. Weihe, N. Wang, R. Deal, and C.E. Boucher. 1998. Creating and restoring wetlands: A whole-ecosystem experiment in self-design. BioScience 48:1019-1030.

Mitsch, W.J. and J.G. Gosselink. 2000. Wetlands, Third Edition. John Wiley & Sons, Inc., New York, NY.

Mitsch, W.J., C.J. Anderson, M.E. Hernandez, A. Altor and L. Zhang. 2004. Net primary productivity of macrophyte communities after ten growing seasons in experimental marshes. p. 75-78. In: W.J. Mitsch, L. Zhang and C.L. Tuttle (eds.) Olentangy River wetland Research Park at The Ohio State University Annual Report 2003. The Ohio State University, Columbus, OH.

Mitsch, W.J. and L. Zhang. 2004. Plant community development after ten growing seasons in two experimental wetland basins. p. 69-73. In: W.J. Mitsch, L. Zhang and C.L. Tuttle (eds.) Olentangy River Wetland Research Park at The Ohio State University Annual Report 2003. The Ohio State University, Columbus, OH.

Mitsch, W.J., N. Wang, L. Zhang, R. Deal, X. Wu and A. Zuwerink. 2005a. Using ecological indicators in a whole-ecosystem wetland experiment. p. 211-235. In S.E. Jørgensen, F-L. Xu, and R. Costanza (eds.) Handbook of Ecological Indicators for Assessment of Ecosystem Health, CRC Press, Boca Raton, FL.

Mitsch, W.J., J.W. Day, Jr., L. Zhang, and R. Lane. 2005b. Nitrate-nitrogen retention by wetlands in the Mississippi River Basin. Ecological Engineering 24: 267-278.

Mitsch, W.J., L. Zhang, C.J. Anderson, A. Altor, and M. Hernandez. 2005c. Creating riverine wetlands: Ecological succession, nutrient retention, and pulsing effects. Ecological Engineering (in press).

104 Nair, V.D., D.A. Graetz, K.R. Reddy, and O.G. Olila. 2001. Soil development in phosphate-mined created wetlands of Florida, USA. Wetlands 21:232-239.

Nairn, R.W. 1996. Biogeochemistry of newly created riparian wetlands: evaluation of water quality changes and soil development. PhD. Dissertation. The Ohio State University, Columbus, OH.

Nairn, R.W. and W.J. Mitsch. 2000. Phosphorus removal in created wetland ponds receiving river overflow. Ecological Engineering 14: 107-126.

Newman, S. and K. Pietro. 2001. Phosphorus storage and release in response to flooding: implications for Everglades stormwater treatment areas. Ecological Engineering 18:23-38.

Newman, S., J. Schuette, J.B. Grace, K. Rutchey, T. Fontaine, K.R. Reddy, and M. Pietrucha. 1998. Factors influencing cattail abundance in the northern Everglades. Aquatic Botany 60:265-280.

Ponnamperuma, F.N. 1972. The chemistry of submerged soils. Advances in Agronomy 24:29-96.

Sanyal, L.K. and L.K. De Datta. 1991. Chemistry of phosphorus transformations in soil. In: B. A. Stewart (ed.) Advances in Soil Science, Volume 16, Springer-Verlag, New York, NY.

Selbo, S.M. and A.A. Snow. 2004. The potential for hybridization between Typha angustifolia and Typha latifolia in a constructed wetland. Aquatic Botany 78: 361- 369.

Shaffer, P.W. and T.L. Ernest. 1999. Distribution of soil organic matter in freshwater emergent/open water wetlands in the Portland, Oregon metropolitan area. Wetlands 19:505-516.

Sommers, L.E. and D.W. Nelson. 1972. Determination of total phosphorus in soils: A rapid perchloric acid digestion proceedings. Soil Science Society of America Proceedings 36: 902-904.

Spieles, D.J. and W.J. Mitsch. 2000a. The effects of season and hydrologic and chemical loading on nitrate retention in constructed wetlands: A comparison of low and high nutrient riverine systems. Ecological Engineering 14: 77-91.

Spieles, D.J. and W.J. Mitsch. 2000b. Macroinvertebrate community structure in high- and low-nutrient constructed wetlands. Wetlands 20: 716-729.

Spieles, D.J. and W.J. Mitsch. 2003. A model of macroinvertebrate trophic structure and oxygen demand in freshwater wetlands. Ecological Modelling 161: 183-194. 105

Systat Software, Inc. 2002. Systat for Windows. Version 10.2. Systat Software, Inc. Richmond, CA.

Thomas, G.W. 1996. Soil pH and soil acidity. p. 475-490. In: Methods of Soil Analysis, Part 3—Chemical Methods. Soil Science Society of America, Madison, WI.

Tiner, R. W. 1999. Wetland Indicators- A Guide to Wetland Identification, Delineation, Classification, and Mapping. Lewis Publishers, Boca Raton, FL.

Vepraskas, M.J. and S.P. Faulkner. 2001. Redox chemistry of hydric soils. p. 85-105. In: J. L. Richardson and M. J. Vepraskas (eds.) Wetland Soils: Genesis, Hydrology, Landscapes and Classification. Lewis Publishers, Boca Raton, FL.

Warncke, D. and J.R. Brown. 1998. Potassium and other basic cations, p. 31-33. In: Recommended Chemical Soil Test Procedures for the North Central Region. NCR Publication No. 221. Missouri Agricultural Experiment Station, Columbia, MO.

Wetzel, R.G. 2001. Limnology: Lake and River Ecosystems, Third Edition. Academic Press, Inc., San Francisco, CA.

Wu, X. and W.J. Mitsch. 1998. Spatial and temporal patterns of algae in newly constructed freshwater wetlands. Wetlands 18: 9-20.

Zedler, J.B. and J.C. Callaway. 1999. Tracking wetland restoration: do mitigation sites follow desired trajectories? Restoration Ecol. 7:69-73.

Zhang, L. and W.J. Mitsch. 2005. Modelling hydrological processes in created wetlands: An integrated system approach. Environmental Modelling & Software 20: 935-946.

106

Figure 4.1 The two experimental wetlands at The Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio pumping system and water control structures.

107 Wetland 1

Wetland 2

Figure 4.2 The 10-m grid and locations used for soil sampling at the ORWRP experimental wetlands in 1993, 1995 and 2004. Shaded areas within the grid map represent approximate location of the deeper, open water (OW) zones.

108 a) b) 12 a 5 c a 10 4 b 8 d a c cd bcd 3 b b 6 c cd 4 2 d 2 1 Perecnt total C (%) Perecnt total 0 0 Percent organic matter (%) matter organic Percent 1993 1995 2004 1993 1995 2004 1993 1995 2004 1993 1995 2004 0-8 cm depth 8-16 cm depth 0-8 cm depth 8-16 cm depth

c) d) a a ab 800 b 16 b

700 c ) 14 -1 ) c

-1 600 12 g c μ

g 500 10

μ d 400 8 300 6 No 200 4 Total P ( Total Data

100 Available P( 2 0 0 1993 1995 2004 1993 1995 2004 1993 1995 2004 1993 1995 2004 0-8 cm depth 8-16 cm depth 0-8 cm depth 8-16 cm depth

Figure 4.3 Comparison of combined mean (±1 SE) for a) percent organic matter (n=32), b) total C (n=18), c) total P (n=18), d) available P (n=28), e) exchangeable Ca (n=28), f) exchangeable K (n=28), g) exchangeable Mg (n=28), and h) soil pH (n=28) at the ORWRP experimental wetlands in 1993, 1995 and 2004 at 0-8 and 8- 16 cm depths. Letters denote differences between years and depths detected at p<0.05 based on Wilcoxon Signed Ranks Test (Bonferroni adjusted).

109

a a 5000 e) 200 f) ) )

-1 4000 b 160 b cd c d -1 c b c c g cd g

3000 μ 120

2000 80

1000 Exch. K ( 40 Exch.( Ca 0 0 1993 1995 2004 1993 1995 2004 1993 1995 2004 1993 1995 2004 0-8 cm depth 8-16 cm depth 0-8 cm depth 8-16 cm depth

g) h) 500 a 7.4 b b c b b a

) a -1 400 7.2

g b b μ 300 7 c c

200 Soil pH 6.8

Exch. Mg ( Mg Exch. 100 6.6

0 6.4 1993 1995 2004 1993 1995 2004 1993 1995 2004 1993 1995 2004 0-8 cm depth 8-16 cm depth 0-8 cm depth 8-16 cm depth

Figure 4.3 (continued).

110 1.00 0.90 a) Wetland 1

) 0.80 1993 0.70 1995 2004 0.60 0.50 0.40 0.30

Frequency (% of total area total of (% Frequency 0.20 0.10 0.00 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 Soil organic matter (%)

1.00 0.90 b) Wetland 2 ) 0.80 1993 0.70 1995 2004 0.60 0.50 0.40 0.30

Frequency (% of total area total of (% Frequency 0.20 0.10 0.00 4.0 5.0 6.0 7.0 8.0 9.0 10.0 11.0 12.0 13.0 14.0 15.0 16.0 Soil organic matter (%)

Figure 4.4 Frequency distribution curves of soil organic matter in 1993, 1995 and 2004 for a) Wetland 1 and b) Wetland 2 based on spatially interpolated data.

111 19 a) 1993

Distance (10 m) Distance

Percent Organic Matter > 13 > 10 > 9 > 8 > 7 > 6 > 5 > 3 0 0 8 8 16 Distance (10 m)

Figure 4.5 Spatial distribution maps of soil organic matter for Wetland 1 and 2 in a) 1993, b) 1995 and c) 2005. Maps for 1993 and 2004 were generated by ordinary point kriging using isotropic variogram models. Maps for 1995 were generated using inverse distance weighing method for spatial interpolation (see text).

112

19 b) 1995

Distance (10 m) Distance

Percent Organic Matter > 13 > 10 > 9 > 8 > 7 > 6 > 5 > 3 0 0 8 8 16 Distance (10 m)

Figure 4.5 (continued)

113

19 c) 2004

Distance (10 m) Distance

Percent Organic Matter > 13 > 10 > 9 > 8 > 7 > 6 > 5 > 3 0 0 8 8 16 Distance (10 m)

Figure 4.5 (continued)

114

Variogram Wetland 1 Wetland 2 characteristic 1993 1995 2004 1993 1995 2004 † nugget (C0) 0.10 -- 0.45 <0.01 -- 4.26

sill (C0 + C) 0.25 -- 5.24 0.49 -- 9.54

range, m (A0) 239 -- 24 152 -- 62

proportion [C/(C0 + C)] 0.59 -- 0.91 0.99 -- 0.57 model‡ Exp -- Sph Sph -- Exp r2 0.98 -- 0.93 0.98 -- 0.91 † No spatial structure detected (total nugget effect) ‡ 'Exp': Exponential and 'Sph': Spherical

Table 4.1 Variogram characteristics for soil percent organic matter concentrations in Wetland 1 and 2 at 0-8 cm depth for 1993, 1995 and 2004.

115

CHAPTER 5

SEDIMENT, CARBON, AND NUTRIENT ACCUMULATION AT TWO 10-

YEAR-OLD CREATED RIVERINE MARSHES

5.1 Abstract

Two 1-ha riverine wetlands at the Olentangy Wetland Research Park in Columbus,

Ohio were constructed in 1993 with a nearly identical geomorphology and have maintained an identical hydrology since their creation. The only initial difference was that one wetland was planted with native macrophytes in 1994 while the other was not.

Sediment and nutrient accumulation was evaluated in May 2004, ten years after the wetlands were created. Higher mean sediment accumulation was detected for the two wetlands in the deeper open water (OW) zones (62.0 ±6 and 74.0 ±5 kg m-2) than in the

emergent (EM) vegetation zones (38.0 ±2 and 39.4 ±3 kg m-2). Directional spatial

structure associated with sediment accumulation was detected in both wetlands and was

attributed to the high accumulation in the OW zones and the gradual decline in

accumulation from inflow to outflow. Despite several years of markedly higher

productivity in Wetland 2, this wetland showed no evidence of higher organic C

116

accumulation however, the dense Typha established during those years may have elicited

greater deposition and reduced re-suspension of sediment in the OW zones. High

accumulations of Ca (2.4 ±0.2 kg m-2 for both wetlands) and inorganic C (730 ±70 and

-2 717 ±49 g m ) in the OW zones of both wetlands suggest that CaCO3 deposition has

remained a critical process where algae productivity has been highest. Annual rates of

sediment and nutrient accumulation for each wetland were lower than those calculated in

previous years and typically fall between ranges seen for newly created wetlands and

natural wetlands.

5.2 Introduction

In open wetland systems such as estuarine and riverine wetlands, sediment flux is

instrumental in shaping edaphic conditions. Wetland factors that have been shown to influence sediment accumulation have included its geomorphology (Hupp and Bazemore

1993, Johnston 2001), hydrology (Mitsch et al. 1995, Pasternack and Brush 1996, Olila et al. 1997, Craft et al. 2002), nutrient load (Richardson and Craft 1993, Brenner et al.

2001), and macrophyte cover (Pasternack and Brush 1996, Horppila and Nurminen 2001).

These factors are even more important in created wetlands where initial soil conditions typically consist of excavated terrestrial soils with characteristics very different to those found in natural wetlands. Because created wetland soils may take decades to develop and because of its ramifications on water quality, more information on wetland factors

and their influence on sediment and nutrient accumulation is needed.

Macrophytes have been shown to influence sediment accretion in two ways. First,

macrophytes and other aquatic vegetation contribute autochthonous organic matter to the

117

wetland surface that augment imported sediment and precipitated minerals. While

allochthonous organic matter is also important, the autochthonous organic matter tends to

be more structurally intact and takes longer to breakdown (Wetzel 2001), thus providing a

longer-term addition to the sediment layer. Macrophytes can also increase sediment

accumulation by altering water column conditions that increase deposition and suppress

re-suspension. Emergent vegetation and detritus increase sedimentation by reducing

water flows, providing substrate for sediments to adhere, and reducing overhead winds

which may cause re-suspension (Braskerud 2001, Horppila and Nurminen 2001).

However, other studies have found that vegetation can have the opposite effect on net

wetland sedimentation by impeding water flow into emergent vegetation zones. It has

been shown that water may preferentially follow through sparsely vegetated areas and

short circuit the vegetated portions of the wetland (Fennessy et al. 1994).

Studies that have evaluated conditions in open wetland systems have shown that newly created wetlands (Fennessy et al. 1994, Braskerud 2001, Harter and Mitsch 2003)

typically report much higher sedimentation rates than those studying older created

wetlands (Craft 2003) or natural wetlands (Johnston 1991, Peterjohn and Correll 1994,

Craft and Casey 2000). Other research has demonstrated that sediment and nutrient

accumulation can vary considerably within wetlands depending upon preferential flow

and proximity to inflows (Reddy et al. 1993, Mitsch et al. 1995). In a conceptual model

developed by Craft (1997), sediment and P accumulation in created estuarine wetlands

tends to occur rapidly in the first few years but as it accumulates, the rate of retention

eventually peaks and declines so that after ten years, they may become comparable to

natural systems. Created wetland removal of N (via denitrification) is eventually dictated

118

by the build-up of C (via organic matter accumulation) and both processes should

increase over a ten year period.

A survey of sediment condition and accumulation was conducted at two experimental

marshes at The Olentangy River Wetland Research Park (ORWRP) ten years after they

were created. These wetlands have been studied extensively over the past ten years with

data collected annually on wetland productivity, hydrology, biogeochemistry and overall

ecology (see Mitsch et al. 1998, 2005, in press, Kang et al. 1998, Koreny et al. 1999,

Spieles and Mitsch 2000a,b, Ahn and Mitsch 2002, Anderson et al. 2003, Selbo and

Snow 2004, and Zhang and Mitsch 2005). Several studies have identified key processes

that have influenced edaphic conditions including metaphyton productivity (Wu and

Mitsch 1998), macrophyte productivity (Mitsch et al. 2005), short-term sedimentation

(Harter and Mitsch 2003), nutrient retention (Nairn and Mitsch 2000, Spieles and Mitsch

2000a) and the co-precipitation of CaCO3 and P (Liptak 2000). Sediment and organic

accretion have been identified as a major contributing factor to soil development in the

two wetlands (Nairn 1996, Harter and Mitsch 2003) and in 2004 the mean depth of the

sediment layer between the two wetlands was 9.3 ±0.4 cm (Anderson et al. 2005). In the

open water (OW) zones, the sediment layer tended to be deepest (8-22 cm), was grey-

black in color, and very homogeneous with very fine particulate matter that was almost gelatinous in structure suggesting the formation of a gyttja layer (Wetzel 2001). The

sediment layer in the emergent vegetation (EM) zones were slightly more cohesive and

heterogeneous with samples containing variable amounts of undistinguishable macrophyte detritial matter, living macrophyte roots/rhizomes, fine mineral/organic sediment, and soil fauna (Anderson et al. 2005).

119

The objectives of this study were to: 1) estimate the 10-yr accumulation of sediment,

C (organic, inorganic and total), N, P, and Ca at the two experimental wetlands; 2)

compare the wetlands to determine if differences in wetland productivity and diversity

over ten years has influenced the variability and accumulation of sediment and nutrients;

3) examine sediment accumulation throughout the wetlands for spatial structure based on non-directional and directional analyses, and to compare the EM and OW zones within each wetland; and 4) compare accumulation rates to previous research at the ORWRP and research reported elsewhere in the literature.

5.3 Methods

5.3.1 Study area

This study was conducted at two created riverine wetlands at the Olentangy River

Wetland Research Park (ORWRP) (Fig. 5.1) on The Ohio State University campus in

Columbus, Ohio (latitude N40.021◦, longitude E83.017◦). The experimental marshes

were excavated in 1993 and flooded on a continuous basis starting in March 1994 with

pumped Olentangy River water. The two wetlands were constructed to be nearly

identical in size (~1 ha) and geomorphology, and pumped to have an identical hydrology.

The only major difference initially was that the western marsh (Wetland 1) was planted

with 2500 individual plants representing 13 native, wetland species while the eastern marsh (Wetland 2) was left unplanted (Mitsch et al. 1998, 2005, Fig. 5.1). Underlying soils in these wetlands were non-hydric alluvial Ross series (classified as a Cumlic

Hapludoll) soils which consisted of silt loam, silt clay, and clay loams (Mcloda 1980).

Prior to construction, specific soil testing at the ORWRP wetland sites revealed

120

subsurface soils (50 to 300 cm below surface) that were grey silty-clay, with low

permeability, and were highly compactable (Nairn 1996). An evaluation of the original

(antecedent) soil surface between 1993 and 2004 found that there was no significant

change in soil bulk density (~1.3 g cm-3) or percent organic matter (~ 5.3%) and that it has remained very distinguishable from the much less cohesive sediment layer

accumulating above it (Anderson et al. 2005). There were some changes observed in the antecedent soil layer including reduced chroma, the export of soluble elements, and the presences of sporadic oxidized rhizospheres; however the vast majority of accreted sediment material has accumulated above the antecedent wetland surface. Consequently, we used this soil boundary as a horizontal marker to estimate sediment accretion throughout both wetlands.

Based on geomorphology, both wetlands have developed two distinct cover zones: 1) a shallow, emergent vegetation (EM) zone and 2) three deeper, open-water (OW) sub- basins spaced longitudinally within each wetland (Fig. 5.2). In general, the EM zones were excavated 30 cm or less below natural grade with the OW zones built as deepwater pockets another 30 cm deeper than the surrounding EM zone. The OW zones have not supported emergent macrophytes; however they often have supported high amounts of

macroalgae and other open-water vegetation. Of the total wetland, the OW basins

represent approximately 30% of the wetland area. Since wetland creation in 1993-94, there were several years when Wetland 2 (the originally unplanted wetland) was substantially more productive and less diverse. Between 1998-2001, mean net above ground primary productivity in Wetland 2 (between 832-1127 g m-2 yr-1) was significantly

higher than Wetland 1 (between 393-729 g m-2 yr-1) (Mitsch et al. 2005, in press). This

121

was caused by the development of a monoculture of Typha angustifolia L. in Wetland 2,

while Wetland 1 remained dominated by a mixed assemblage of introduced yet less

productive species. Based on a community diversity index, Wetland 1 has had higher species diversity compared to Wetland 2 (Mitsch et al. 2005, in press). Other than vegetation, the two wetlands have been constructed, managed, and pumped identically since 1994.

5.3.2 Soil sampling and analyses

Sediment sampling occurred in May 2004 (10 years and 2 mos. after water was first introduced to these basins by pumping) using a 10-m grid system (Fig. 5.2) established in

1993. Cores were used to measure sediment depth (measuring from the surface to the antecedent soil surface) at a total of 127 grid points providing an even spatial distribution throughout both wetlands (Fig. 5.2). Sediment cores were extracted using a 10-cm diameter steel corer at 16-cm depth increments until the antecedent clay-soil layer was reached. To analyze sediment physiochemical characteristics, a total of 69 sediment samples were collected for physiochemical analyses. The top 8 cm of sediment were collected at all grid points where sediment depth was >8 cm deep. Each core was visually

inspected, split vertically into two equally sized sections, and placed in separate water-

tight freezer bags. Collected sediment samples were kept in an iced cooler until being

returned to the laboratory where they were stored at 4 ◦C.

At the laboratory, each sediment sample was homogenized by hand. One section

from each sampling point was oven-dried at 105ûC for determination of bulk density.

The second section of each soil sample was kept in its field-moist, natural condition and a

122

30-g subsample of each sample was air-dried at room temperature, ground using a pestle

and mortar, and passed through a 2-mm sieve. Duplicate subsamples (approximately 10 g each) were placed in a crucible, oven-dried at 60 ûC overnight, weighed, and ignited in a muffle furnace at 550ûC for 1 hour. Ten subsamples were further analyzed for textural conditions using the hydrometer method (Gee and Bauder 1986). Using the field-moist sections, 47 100-g subsamples were air-dried at room temperature, ground using a pestle and mortar, passed through a 2 mm sieve, and. analyzed for total N and C by combustion

(AOAC 1989, ISO 1995). From this group, 21 samples were further analyzed for total P and Ca by digestion with HClO4/HNO3 followed by inductively coupled plasma emission spectrometry (Sommers and Nelson 1972). A total of 22 samples were analyzed for inorganic C (USEPA 2005). Percent organic C was calculated based on the difference of total C and inorganic C and a regression was used to estimate percent organic C for all samples based on its percent organic matter (Konen et al. 2002). For all chemical analyses, samples were selected to provide an even spatial distribution and proportionate sampling between cover zones.

A weighted average of each nutrient concentration was calculated for both wetlands based on the percent surface area of each cover zone. The samples analyzed for nutrient concentrations were also used to characterize sediment conditions within each respective wetland cover zone. Mean sediment and nutrient content (g cm-3) were calculated for

each individual wetland cover zone using sampled concentrations and bulk densities.

Mean content was multiplied with sediment depth to estimate sediment and nutrient

accumulation (converted to g m-2) at each grid point measured for depth. Mean

accumulation for each wetland cover zone was calculated and a weighted average was

123

calculated for the each entire wetland. Mean sediment and nutrient accumulations for

each wetland were used to estimate annual accumulation rates (kg yr-1) based on a

wetland age of 10.2 years.

5.3.3 Statistical analyses

The mean and standard error of sediment/nutrient concentrations and accumulations

were determined for each wetland cover zone. Given the non-independent and potentially

autocorrelated nature of the grid-sediment data, standard parametric statistical analyses

were considered inappropriate and not used. To evaluate if wetland differences in 10-yr

macrophyte diversity influenced sediment and organic C accumulation sample variance,

Wetland 1 and 2 were compared using Levene’s Test for equal variances, with P<0.05

considered a significant difference. Data was transformed to meet a normal distribution

(based on Ryan-Joiner test) if necessary. Levene’s test for equal variances, the Ryan-

Joiner test for normality, and liner regression of soil organic matter and organic C were

conducted using Minitab Release 14.0 (Minitab, Inc. 2003).

Sediment accumulation data at Wetland 1 and 2 were evaluated using geostatistical

analyses. Spatial autocorrelation was assessed based on the semivariance of paired

groups of data points within each wetland (each wetland was analyzed separately) (Isaaks

and Srivastava 1989). Isotropic and anisotropic spherical variograms were used to

analyze semivariance based on 10m interval classes over a 60m lag distance with h-

scatterplots used to detect for outliers or aberrant data that may have had excessive

influence on model parameters (Isaaks and Srivastava 1989). Variograms consist of a

graphical output in which the semivariance is measured at increasingly further interval

124

distances. Characteristics of the variogram graph include 1) the nugget variance (C0) which is the experimental variance unaccounted for by the spatial model, 2) the sill (C0 +

C) which is the total variance as measured at the asymptote of the variogram, and 3) the

range (A0) which is the spatial distance at which autocorrelation is detected. When

autocorrelation occurs, the level of variance between interval classes increases and

eventually reaches an asymptote (the sill, C0 + C) and levels off, representing the extent

of autocorrelation (Isaaks and Srivastava 1989). Anisotropic variograms were analyzed at

0◦, 45◦, 90◦ and 135◦ directions to detect if spatial patterns were dependent upon specific

directions. Evidence of anisotropic spatial structure usually indicates an underlying

environmental gradient and these analyses were used specifically to determine if spatial

patterns may occur longitudinally from wetland inflow to outflow (Fig. 5.1).

To compare spatial structure between Wetland 1 and 2, isotropic and anisotropic

correlograms were also prepared using Moran’s I (Perry et al. 2002). Moran’s I is a

conventional measure of autocorrelation and it calculates a correlation statistic between

the increasing distance intervals. The correlation statistic is between +1 and -1 depending

on the degree and direction of the correlation and can be interpreted similar to a Pearson’s

Product Moment correlation. Correlograms were conducted concurrently with

variograms using the same distance interval classes (10 m) and lag distance (60 m).

Likewise, anisotropic correlograms were prepared to analyze Moran’s I values at 0◦, 45◦,

90◦ and 135◦ directions.

Spatial structure detected through variogram analyses were used to conduct kriging

analyses, which is an unbiased procedure that uses the modeled spatial relationship to

interpolate values between data points (Isaaks and Srivastava 1989). The interpolated

125

data from each kriging analysis were used to compare frequency distributions of sediment

accumulation in Wetland 1 (n=79) and 2 (n=92) and to prepare kriging maps that

illustrated the range and location of sediment accumulation. All geostatistical analyses

were conducted using GS+ Software (Version 7.0) (Gamma Design Software 2004).

5.4 Results

5.4.1 Sediment characteristics and accumulation

Mean bulk density of the EM zone sediment in Wetland 1 and 2 was 0.55 ±0.02 and

0.57 ±0.05 g cm-3, respectively (Table 5.1). Mean bulk density of the OW sediment in

Wetland 1 and 2 was 0.44 ±0.03 and 0.48 ±0.04 g cm-3, respectively (Table 5.1;

Appendix C, Table C.1). Sediment was silt loam in texture with the percentage of sand, silt, and clay estimated at 37 ±3, 50 ±2, and 13±2%, respectively, with OW samples having slightly more clay than sand (Appendix C, Table C.4). Otherwise, textural conditions showed no patterns between wetlands and cover zones. Nutrient

concentrations in the sediment layer varied between zones within wetlands; however, no

substantial differences were noted between total wetland estimates (Table 5.1).

Compared to their OW zones, mean organic C concentrations were higher in the EM

zones of Wetland 1 and Wetland 2 (24 and 35% higher, respectively). Substantial

differences between zones were also noted at Wetland 1 and 2 for total P and Ca (Table

5.1).

Based on measured sediment depths above the antecedent soil layer, accumulation

was extensive in Wetland 1 and 2 with a mean of 45.7 ±3.5 and 49.8 ±3.1 kg m-2,

126

respectively over 10 years or an average of about 4.5 – 4.9 kg m-2 yr-1. The highest

amount of

sediment accumulation was detected in the OW zones of Wetland 2 which had an average

of 74.0 ±4.6 kg m-2 (Fig. 5.3a). This was approximately 19% greater than sediment

accumulation estimated in the OW zones of Wetland 1 however total wetland

accumulation was comparable (Fig. 5.3a). Results from Levene’s Test for equal

variances indicated that there was no significant difference (P>0.05) in sample variance between wetlands. Examining the frequency distribution of sediment accumulation based on kriging analysis (Fig. 5.3b, see variograms results below), both wetlands had

significant overlap, however Wetland 2 had a greater frequency of high range sediment

accumulation (>60 kg m-2) because of the higher accumulation in its OW zones.

5.4.2 Spatial patterns of sediment accumulation

In Wetland 1, no spatial structure associated with sediment accumulation was

detected based on prepared isotropic variograms and correlograms. Strong spatial

◦ structure was detected at the 0 direction based on an anisotropic variogram (C0=61.2,

C0+C=480.3, A0=5.1) and correlogram (Fig. 5.4a). Reviewing correlogram results, a

Moran’s I value of 0.63 was determined for grid pairs at the first lag interval and steadily decreased over the 60 m lag distance. None of the other anisotropic directions provided

evidence of spatial structure. In Wetland 2, evidence of moderate spatial structure was

detected in the isotropic correlogram along with comparable anisotropic structure at 0◦

◦ ◦ and 135 directions (Fig. 5.4b). However, the 135 variogram (C0=160.8, C0+C=616.0,

A0=4.8) was the only one that corroborated with the correlogram results and therefore

127

was used for kriging analyses. In the 135◦ correlogram, a Moran’s I value of 0.30 was

detected at the first interval, immediately dropped below 0 at the second lag interval, and stayed near 0 throughout the remaining lag distance. All anistotropic variograms and correlograms were re-analyzed after adjusting the direction at 5◦ intervals, however, none

of the adjusted variograms improved upon the results reported.

Kriging maps of Wetland 1 and 2 showed the range and locations of sediment

accumulation throughout the wetlands (Fig. 5.5). Kriging analysis tends to suppress the

range of interpolated values; therefore extreme high and low measurements tend to be

less represented by this procedure. Nevertheless, in Wetland 2, there was a clear

distinction of higher accumulation occurring within all three OW sub-basins compared to

its surrounding EM zone (Fig. 5.5). In Wetland 1, only the middle sub-basin showed the

same distinction. Both Wetland 1 and 2 showed highly variable conditions in the EM

zones with accumulations ranging from 13.9 to 70.9 kg m-2.

5.4.3 Nutrient accumulation

5.4.3.1 Carbon

No substantial differences in total C accumulation (Fig. 5.6a) were observed between

wetlands, however there was 17% more in the OW zones of Wetland 2 (2.8 ±0.2 kg m-2)

compared to Wetland 1 (2.4 ±0.2 kg m-2). This was primarily due to the accumulation of

organic C which represented the majority of detected C accumulation and had similar

proportions between wetlands and cover zones as did total C (Fig. 5.6b). Levene’s Test

for equal variances indicated no significant difference (P>0.05) in organic C sample

variance between wetlands. Both wetlands had significant amounts of inorganic C in the

128

OW zones sediment (Fig. 5.6c). Inorganic C represented 30 and 26% of the total C

accumulated in the OW zones of Wetland 1 and 2, respectively. Based on the mean

wetland accumulation total, annual rates of total, organic and inorganic C were prepared

(Table 5.2). Wetland 2 had higher rates of all C constituents except for inorganic C.

5.4.3.2 Nitrogen, phosphorus and calcium

Total N accumulation was nearly identical in Wetland 1 and 2, with slightly higher

accumulations detected in the OW zones (Fig. 5.7a). Highest N accumulation was

detected in the OW zone of Wetland 2 (228 ±14 g m-2). P accumulation was also highest

in the OW zones of Wetlands 2 (58 ±4 g m-2) and was 41% higher than the OW zones of

Wetland 1 (41 ±4 g m-2) (Fig. 5.7b). However, because P accumulation in the EM zone

was higher in Wetland 1, differences in total wetland P accumulation were negligible.

Total Ca accumulation was consistent between wetlands but showed extreme intra-

wetland variation (Fig. 5.7c). In Wetland 1 and 2, total Ca accumulation was 13 and 11

times higher, respectively, in the OW zones compared to respective EM zones (Fig. 5.7c;

Appendix C, Table C.5). Annual accumulation rates and net accumulation of N, P and

Ca were similar between wetlands (Table 5.2).

5.5 Discussion

5.5.1 Sediment accumulation and spatial patterns

Based on whole-wetland estimates, there was no evidence that the higher 10-yr productivity in the naturally colonizing Wetland 2 resulted in a greater accumulation of sediment. We had expected that wetland to have more sediment accumulation primarily

129

through greater accretion of autochthonous organic matter, but accumulations of sediment and organic C were similar between the EM zones of both wetlands (Fig. 5.3a and 6b).

However sediment accumulation was 19% higher in the OW zones of Wetland 2 compared to Wetland 1, and the sediment kriging maps clearly showed that the OW zones of Wetland 2 had accumulated more than Wetland 1. It is uncertain exactly why this difference occurred but it is possible that vegetation may have played a role. Between

1998 and 2001, Wetland 2 was dominated by dense stands of T. angustifolia that stood well above 2 m in height with mean shoot densities >32 m-2 (Selbo and Snow 2004).

Vegetation communities in Wetland 1 tended to be more diverse but lower-growing and with less productive macrophytes. It is possible that the dense Typha stands surrounding the OW zones of Wetland 2 were more effectively sheltered these areas and increased sediment deposition while suppressing re-suspension. Other studies have found evidence that plant density and height can influence sediment deposition rates (Horppila and

Nurminen 2001, Darke and Megonigal 2003). This would explain why accumulation rates were highest at the inflow sub-basin of Wetland 2 and were higher overall compared to

Wetland 1. The highest sediment accumulation in the planted wetland occurred at the middle sub-basin after it had been transported a considerable distance from the inflow.

Both wetlands demonstrated spatial structure associated with the accumulation of sediment. The 0◦ anisotropic results in the planted Wetland 1 suggest that spatial structure was occurring primarily in a north-to-south direction. This effect was caused in part by the gradual decrease in overall accumulation from inflow to outflow, but also the influence from the middle OW sub-basin which is elongated in a north-south direction

(Fig. 5.2). This sub-basin was the only one that accumulated sediment in stark contrast to

130

the surrounding EM zones and its concentric nature within the wetland likely contributed

to the 0◦ spatial structure detected (Fig. 5.5a). This was in contrast to the naturally

colonized Wetland 2 which exhibited more moderate spatial structure but at isotropic and

anisotropic (0◦ and 135◦) directions (Fig. 5.5b). This was caused by the same decrease in

accumulation from inflow to outflow seen in Wetland 1, but with a greater influence by

all three OW sub-basins. As indicated in the sediment kriging maps (Fig. 5.5), the greatest contrast between the EM and OW zones in Wetland 2 occurred at the inflow

(north) section of the wetland which was analyzed most extensively using the 135◦ anisotropic analyses (Fig. 5.4). While this direction produced correlogram results similar to other directions, the 135◦ variogram was the only one that detected spatial structure. In

both wetlands, the variograms and correlograms that detected spatial structure were using

directional analyses that most extensively intercepted the OW sub-basin with the highest

sediment accrual. Therefore, we conclude that these wetland features had a prevailing

influence on sediment spatial structure.

5.5.2 Longitudinal patterns

Both wetlands showed a general longitudinal decrease in sediment accumulation.

This is a common occurrence in flow-through wetlands where sedimentation rates are

highest near the inflow source and decrease as they continue through the system

(Fennessy et al. 1994, Braskerud 2001). In the case of both Wetland 1 and 2, the southern half of each wetland had EM zones with sediment accumulation at <40 kg (with some inclusions) compared to the northern half that was primarily >40 kg. Both wetlands also had some of the least sediment accumulation occur in the EM zones along the concave

131

mid-section of each wetland (Fig. 5.4). This area of both wetlands represents some of the

shallowest sections and consists of a gradual elevation relief from the wetland edge to the

mid OW sub-basin. The low sediment accumulation in these sections and high

accumulation in the adjacent OW zone suggests that water is preferentially flowing

through the open water area and by-passing this shallow emergent area. It was also noted

that the OW zones in Wetland 2 also followed a decreasing accumulation gradient from

inflow to outflow, unlike Wetland 1. Again, this may be the result of several years of

dense Typha that may have promoted earlier sedimentation (closer to the inflow) and

reduced re-suspension.

Other factors also influenced sediment accumulation. Local topographic conditions

were shown to have an effect as sediment accretion was often minimal near the crest of

the OW sub-basins (Anderson et al. 2005). Without the benefit of rooted macrophyte

cover, these areas appear to be highly susceptible to sediment transport either southward

towards the outflow or sloughing downward into the deeper sections of the OW sub-

basin. Other factors influencing sediment distribution are more difficult to account for.

Faunal species such as beaver (Castor canadensis) and muskrats (Ondatra zibethicus) have been active in the wetlands during portions of their history and have likely contributed to the patchiness of sediment seen.

5.5.3 Sedimentation rates

When averaged for the entire wetland, sediment accumulation rates for Wetland 1 and

2 were between those reported for newly created wetlands and those seen in natural wetlands. An investigation of short-term sedimentation at Wetland 1 and 2 (between

132

1996 and 1997) using horizon markers showed an average sedimentation rate of 36 kg m-

2 yr-1 (Harter and Mitsch 2003). High sediment accumulation rates have also been reported in other newly created wetlands. At the Des Plaines River Wetland Demonstration Project

north of Chicago, Illinois, Fennessy et al. (1994) reported rates between 5.9 – 12.8 kg m-2

yr-1 during its first year and 1.2 – 4.2 kg m-2 yr-1 during its second year. Braskerud (2001)

reported rates between 15 – 75 kg m-2 yr-1 for four small newly-created wetlands in an

agricultural watershed in Norway, and Craft (1997) reported rates between 21 – 36 kg m-2

yr-1 for newly constructed estuarine marshes in North Carolina. Our study found overall

annual sedimentation rates (4.5 – 4.9 kg m-2 yr-1) that were much lower than those found

by the Harter and Mitsch (2003) horizon marker study suggesting that the wetlands are no

longer retaining sediment at the pace it was in the first few years after it was constructed.

This was supported by long term water quality data that was collected daily from the

inflow and outflow of both wetlands since they were flooded (Mitsch and Zhang 2004).

In the three years leading up to this sediment survey, both wetlands showed a decrease in

turbidity abatement between inflow and outflow (an indicator of sediment retention), and

by 2001, both wetlands had become sediment sources (Mitsch et al. 2004). This

development has also influenced P retention and as of 2003 both wetlands have exported

more P than retained (Mitsch et al. 2004).

5.5.4 Nutrient accumulation

5.5.4.1 Carbon

133

The lack of any substantial differences in C accumulation between Wetland 1 and 2

was somewhat unexpected given the higher 10-yr productivity in Wetland 2. However

since 2001, the planted Wetland 1 has been the more productive wetland (mean net

annual aboveground primary productivity after 2001 was 561 ±77 in Wetland 1 and 356

±59 in Wetland 2) and the higher deposition of recent organic matter may explain why

the two wetlands are now similar in total and organic C. Evidence of the higher

proportion of recent organic deposition in Wetland 1 is also reflected by the higher

concentrations of total N detected in Wetland 1 (Table 5.1).

The higher accumulation of inorganic C in the OW zones of each wetland is an

indication that CaCO3 precipitation has remained a critical process in sediment condition.

Macrophyte shading has effectively precluded this process in the EM zones and therefore

accumulation was substantially less. Higher accumulation of inorganic C in the EM zone

of Wetland 1 (compared to Wetland 2) suggests that overall algal productivity may have

also been higher. This would be expected given several years in which dense Typha

stands in Wetland 2 likely precluded algae growth more so than vegetation communities

in Wetland 1.

C accumulation rates at the ORWRP wetlands were within the range seen in the

literature. In an evaluation of four created estuarine marshes in North Carolina (1-15 yrs

old), Craft (1997) found the accumulation of organic C at an average rate of 80 g m-2 yr-1.

This rate was also determined for reference natural wetlands used for comparison. Along a Typha gradient in the anthropogenically influenced section of the Everglades, rates for

C accumulation have been documented between 86 – 387 g m-2 yr-1 (Reddy et al. 1993)

and soil organic matter at 492 – 1160 g m-2 yr-1 (Davis 1991).

134

5.5.4.2 Nitrogen, phosphorus and calcium

Wetland removal/retention of N is typically dependent upon organic matter

accumulation as an energy source for denitrifying bacteria and for retaining organically

bound N. At the two experimental wetlands, N accumulated in proportions similar to organic C and minimal differences between wetlands were detected (16.2 and 16.6 g m-2

yr-1 for Wetland 1 and 2, respectively). These rates are above average for created

wetlands but are comparable to other studies where high nutrient loads cause increased

primary productivity and deposition of organic matter (Craft 1997). Armentano and

Woodwell (1975) found N accumulation was 14 g m-2 yr-1 in a marsh on Long Island, NY

and Reddy et al. (1993) found that Typha dominated areas of the Everglades most

affected by anthropogenic nutrient loads had N accumulation rates of 11 – 24 g m-2 yr-1,

compared to unaffected areas that only averaged 5 g m-2 yr-1. Nutrient loading and

organic accretion likely explains the high N storage in the ORWRP sediment as these

wetlands are pumped at a fairly high rate (>22 m yr-1 in 2003, Zhang and Mitsch 2004)

with river water that is often very high in nitrate concentration (mean NO3 + NO2 inflow concentration for 2003 was 4.1 ±0.3 mg N L-1, Mitsch et al. 2004).

As seen in other riverine wetlands, P accumulation in the sediment layer can be

attributed to a combination of organic matter build-up, deposition of P bound sediment,

and biochemical processes such as sorption and precipitation of dissolved P (Johnston

1991, Axt and Walbridge 1999). We expect that all these processes have occurred at the

ORWRP wetlands. Water column productivity in both wetlands has remained high since

the wetlands were flooded and the co-precipitation of CaCO3 and P has been found to be

135

a highly significant P-retention process in these wetlands (Liptak 2000). Like inorganic

C, the accumulation of Ca was exceptionally high in the wetland OW zones compared to

the EM zones because of the precipitation of CaCO3. A nearly perfect correlation

(r2=0.99) was detected between Ca and inorganic C suggesting that carbonates are the

primary form of sediment Ca fractions. Because P readily adsorbs to CaCO3, this process

has likely contributed to significant P retention within both wetlands. Between 1994 and

1998, Liptak (2000) found precipitated calcite had accumulated at 0.45 kg m-2 yr-1and

was one of the primary mechanisms for P sorption. Likewise, in a 10-yr old created and

reference coastal marsh in North Carolina, Craft (1997) found that large amounts of Ca in

the soil (159 - 516 g m-2) contributed to extensive P sorption (6.1 - 8.4 g m-2 yr-1).

Calcite and other forms of CaCO3 have been attributed to the sorption of other elements beside P including NO3 (Jurinak and Griffin 1992) and K (Galvez-Cloutier and

Dube, 1998). Total S concentrations were found to 3 to 4 times higher in the OW zones

of the ORWRP wetlands compared to the EM zones (Appendix C; Table C.5). There has

been minimal research regarding the potential sorption of S and CaCO3, however our

evidence strongly suggests that this may be occurring. Although S constituents were not

2- determined, Vepraskas and Faulkner (2001) reported that SO4 can be retained by the

3- 3- same adsorption processes that affect PO4 , although PO4 tends to displace it when both

are available. The adsorptive capacity of CaCO3 may also contribute to the high N

accumulations rates observed in both wetlands.

Based on the conceptual model by Craft (1997), P removal is dominated by sedimentation and sorption/precipitation processes during its first three years but after ten years, removal declines significantly and becomes increasingly dependent on organic

136

matter accumulation. This pattern appears to be occurring at the ORWRP wetlands. P

accumulation rates for the two ORWRP wetlands were generally less than those seen at

newly created wetlands and were comparable to natural wetlands where rapid accretion

occurs. Previous estimates of P retention at the ORWRP wetlands have been conducted

by analyzing changes in surface water concentrations. Through water quality analyses,

Nairn and Mitsch (2000) estimated that in the first two years, Wetland 1 and 2 retained

6.7 and 7.5 g m-2 yr-1, respectively although water quality analyses can sometimes

overestimate retention compared to actual sedimentation rates (Mitsch et al. 1995). After

ten years, P accumulation rates in Wetland 1 and 2 have decreased to 3.3 and 3.5 g m-2 yr-

1, respectively. Other researchers have found comparable rates in natural open wetland systems. DeLaune and Patrick (1980) found P accumulated at 2.3 g m-2 yr-1 in a

Louisiana salt marsh and Cooper and Gilliam (1987) found P accumulation was 4.3 g m-2

yr-1 for riparian areas in North Carolina. Highly productive Typha invaded areas in the

Everglades were found to have an elevated rate of P accumulation (0.54 – 1.14 g m-2 yr-1) compared to areas of still dominated by Cladium (0.11 – 0.25 g m-2 yr-1, Reddy et al.

1993 and 0.4 g m-2 yr-1, Richardson and Craft 1993).

5.6 Conclusions

After ten years, sediment accumulations within the two experimental marshes were

highly variable, but mean accumulation rates were generally between those seen in newly

created and natural open-wetlands. The decrease in sediment accumulation rates from

those reported for 1996-1997 by Harter and Mitsch (2001) and water quality data from

the last three years, indicates that these wetlands may no longer function as significant

137

sinks of sediment and P. Despite several years of markedly higher productivity in

Wetland 2, this wetland showed no evidence of higher organic C accumulation.

However, the dense Typha established during those years may have elicited greater

sediment deposition and reduced re-suspension in the OW zones. Sediment accumulation

in both wetlands showed anisotropic spatial structure that was caused in part because by

the decrease in accumulation from inflow to outflow, but more so by the influence of high

accumulation in OW sub-basins. Nutrient accumulation was consistent between

wetlands, however high intra-wetland variation occurred for inorganic C, P and Ca

indicating that the co-precipitation of CaCO3 in the OW zones is still a primary factor on

sediment condition.

5.7 Acknowledgements

Funding for this project was provided through an Ohio Agricultural Research &

Development Center Graduate Research Enhancement Grant, the USDA (Grant No.

2002-35102-13518) and by support from the School of Natural Resources at The Ohio

State University. Anne Altor assisted with soil texture analyses.

5.8 Literature Cited

Anderson, C.J., W.J. Mitsch, and R.W. Nairn. 2005. Temporal and spatial development of surface soil conditions at two created riverine marshes. Journal of Environmental Quality 34(6): 2072-2081.

Anderson, K.L., K.A. Wheeler, J.A. Robinson, and O.H. Tuovinen. 2002. Atrazine mineralization potential in two wetlands. Water Research 36: 4785-4794.

AOAC Official Methods of Analysis. 1989. Method 990.03. Protein(crude) in Animal Feed Combustion Method(Dumas method). 17th edition 2002. Reference: JAOAC 72, 770 (1989).

138

Armentano, T. V. and G. M. Woodwell. 1975. Sedimentation rates in a Long Island marsh determined by 210Pb dating. Limnol. Oceanogr. 20:452-455.

Axt, J. R. and M. R. Walbridge. 1999. Phosphate removal capacity of palustrine forested wetlands and adjacent uplands in Virginia. Soil Sci. Soc. Am. J. 63:1019-1031.

Braskerud, B. C. 2001. The influence of vegetation on sedimentation and resuspension of soil particles in small constructed wetlands. J. Environ. Qual. 30:1447-1457.

Brenner, M., C. L. Schelske, and L. W. Keenan. 2001. Historical rates of sediment and nutrient accumulation in marshes of the Upper St. Johns River Basin, Florida, U.S.A. Journal of Paleolimnology 23:241-257.

Brueske, C. C. and G. W. Barrett. 1994. Effects of vegetation and hydrologic load on sedimentation patterns in experimental wetland ecosystems. Ecological Engineering 3:429-447.

Cooper, J. R. and J. W. Gilliam. 1987. Phosphorus redistribution from cultivated fields into riparian areas. Soil Sci. Soc. Am. J. 51:1600-1604.

Craft, C. B. 1997. Dynamics of nitrogen and phosphorus retention during wetland ecosystem succession. Wetlands Ecology and Management 4:177-187.

Craft, C. B. and W. P. Casey. 2000. Sediment and nutrient accumulation in floodplain and depressional freshwater wetlands of Georgia, USA. Wetlands 20: 323-332.

Craft, C. B., S. Broome, and C. Carlton. 2002. Fifteen years of vegetation and soil development after brackish-water marsh creation. Restoration Ecology 10: 248-258.

Darke, A. K. and J. P. Megonigal. 2003. Control of sediment deposition rates in two mid-Atlantic Coast tidal freshwater wetlands. Estuarine, Coastal and Shelf Science 57: 255-268.

Davis, S. M. 1991. Growth, decomposition, and nutrient retention of Cladium jamaicense Crantz and Typha domingensis Pers in Florida Everglades. Aquat. Bot. 40:203-224.

139

DeLaune, R. D. and W. H. Patrick, Jr. 1980. Rate of sedimentation and its role in nutrient cycling in a Louisiana salt marsh. p. 401-412. In P. Hamilton and K. B. Macdonald (eds.) Estuarine and wetland Processes with Emphasis on Modeling. Plenum Publishing, New York, NY.

Fennessy, M. S., C. C. Brueske, and W. J. Mitsch. 1994. Sediment deposition patterns in restored freshwater wetlands using sediment traps. Ecological Engineering 3:409- 428.

Galvez-Cloutier, R. and J. S. Dube. 1998. An evaluation of freshwater sediments contamination: the Lachine Canal sediments case, Montreal, Canada. I. Quality assessment. Water, Air and Soil Poll. 102:259-279.

Gee, G.W. and J.W. Bauder. 1986. Particle-size analysis. p. 383–411. In A. Klute (ed.) Methods of soil analysis. Part 1. 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI.

Harter, S. K. and W. J. Mitsch. 2003. Patterns of short-term sedimentation in a freshwater created marsh. J. Environ. Qual. 32:325-334.

Horppila, J. and L. Nurminen. 2001. The effect of an emergent macrophyte (Typha angustifolia) on sediment resuspension in a shallow north temperate lake. Freshwater Biology 46:1447-1455.

Hupp, C. R. and D. E. Bazemore. 1993. Temporal and spatial patterns of wetland sedimentation, West Tennessee. Journal of Hydrology 141:179-196.

International Standard Organization, ISO. 1995. Soil quality – Determination of organic and total carbon after dry combustion (elementary analysis). International Organization for Standardization 10694:1995(E). Geneva, Switzerland.

Isaaks, E.H. and R.M. Srivastava. 1989. An Introduction to Applied Geostatistics. Oxford University Press, Inc. New York, NY.

Johnston, C. A. 1991. Sediment and nutrient retention by freshwater wetlands: effects on surface water quality. Critical Reviews in Environmental Control 21:491-565.

Johnston, C. A., S. D. Bridgham, J. P. Schubauer-Berigan. 2001. Nutrient dynamics in relation to geomorphology of riverine wetlands. Soil Sci. Soc. Am. J. 65:557-577.

Jurinak, J. J. and R. A. Griffin. 1992. Nitrate ion adsorption by calcium carbonate. Soil Sci. 113: 130-135.

140

Kang, H., C. Freeman, D. Lee, and W.J. Mitsch. 1998. Enzyme activities in constructed wetlands: Implication for water quality amelioration. Hydrobiologia 368:231-235.

Konen, M. E., P. M. Jacobs, C. L. Burras, B. J. Talaga, and J. A. Mason. 2002. Equations for predicting soil organic carbon using loss-on-ignition for north central U.S. soils. Soil Sci. Soc. Am. J. 66:1878-1881.

Koreny, J.S., W.J. Mitsch, E.S. Bair, and X. Wu. 1999. Regional and local hydrology of a created riparian wetland system. Wetlands 19:182-193.

Liptak, M. 2000. Water column productivity, calcite precipitation, and phosphorus dynamics in freshwater marshes. PhD. Dissertation. The Ohio State University, Columbus, OH, USA.

Mcloda, N. A. and R. J. Parkinson, 1980. Soil Survey of Franklin County, Ohio. USDA- SCS. US Government Printing Office, Washington, DC, USA.

Mitsch, W. J., J. K. Cronk, X. Wu, and R. W. Nairn. 1995. Phosphorus retention in constructed freshwater riparian marshes. Ecological Applications 5: 830-845.

Mitsch, W.J., X. Wu, R.W. Nairn, P.E. Weihe, N. Wang, R. Deal, and C.E. Boucher. 1998. Creating and restoring wetlands: A whole-ecosystem experiment in self-design. BioScience 48:1019-1030.

Mitsch, W. J., C. J. Anderson, M. E. Hernandez, A. Altor and L. Zhang. 2004a. Net primary productivity of macrophyte communities after ten growing seasons in experimental marshes. p. 75-78. In W. J. Mitsch, L. Zhang and C. L. Tuttle (eds.) Olentangy River wetland Research Park at The Ohio State University Annual Report 2003. The Ohio State University, Columbus, OH, USA.

Mitsch, W. J., L. Zhang, N. Dillon, and D. Fink. 2004b. Biogeochemical patterns of created riparian wetlands: tenth-year results (2003). p. 59-68. In W. J. Mitsch, L. Zhang and C. L. Tuttle (eds.) Olentangy River wetland Research Park at The Ohio State University Annual Report 2003. The Ohio State University, Columbus, OH, USA.

Mitsch, W. J., N. Wang, L. Zhang, R. Deal, X. Wu, and A. Zuwerink. 2005. Using ecological indicators in a whole-ecosystem wetland experiment p. 211-236. In S. E. Jorgensen, R. Costanza and F. L. Xu (eds.) Handbook of Ecological Indicators for Assessment for Ecosystem Health. Lewis Publishers, CRC Press, Inc., Boca Raton, FL, USA.

Mitsch, W.J., L. Zhang, C.J. Anderson, A. Altor, and M. Hernandez. Creating riverine wetlands: Ecological succession, nutrient retention, and pulsing effects. Ecological Engineering. (In Press).

141

Nairn, R. W., 1996. Biogeochemistry of newly created riparian wetlands: evaluation of water quality changes and soil development. PhD. Dissertation. The Ohio State University, Columbus, OH, USA.

Nairn, R.W. and W.J. Mitsch. 2000. Phosphorus removal in created wetland ponds receiving river overflow. Ecological Engineering 14: 107-126.

Olila, O. G., K. R. Reddy, and D. L. Stites. 1997. Influence of draining on soil phosphorus forms and distribution in a constructed wetland. Ecological Engineering 9:157-169.

Pasternack, G. B. and G. S. Brush. 1996. Sedimentation cycles in a river-mouth tidal freshwater marsh. Estuaries 21:407-415.

Perry, J. N., A. M. Liebhold, M. S. Rosenberg, J. Dungan, M. Miriti, A. Jakomulska and S. Citron-Pousty. 2002. Illustrations and guidelines for selecting statistical methods for quantifying spatial patterns in ecological data. Ecography 25: 578-600.

Peterjohn, W. T. and D. L. Correll. 1984. Nutrient dynamics in an agricultural watershed; observations on the role of a riparian forest. Ecol. 65: 1466-1475.

Reddy, K. R., R. D. DeLaune, W. F. DeBusk and M. S. Koch. 1993. Long-term nutrient accumulation rates in the Everglades. Soil Sci. Soc. Am. J. 57:1147-1155.

Richardson, C. J. and C. B. Craft. 1993. Effective phosphorus retention in wetlands: fact or fiction? p. 271-282. In G. A. Moshiri (ed.) Constructed Wetlands for Water Quality Improvement. Lewis Publishers, CRC Press, Inc., Boca Raton, FL, USA.

Selbo, S.M. and A.A. Snow. 2004. The potential for hybridization between Typha angustifolia and Typha latifolia in a constructed wetland. Aquatic Botany 78: 361- 369.

Sommers, L. E. and D. W. Nelson. 1972. Determination of Total Phosphorus in Soils: A Rapid Perchloric Acid Digestion Procedure. Soil Science Society of America Proceedings. Vol. 36, 6: p.902-904. Madison, WI.

Spieles, D.J. and W.J. Mitsch. 2000b. Macroinvertebrate community structure in high- and low-nutrient constructed wetlands. Wetlands 20: 716-729.

US EPA Test Methods for Evaluating Solid Wastes. 2005. Chapter 5 Miscellaneous Test Methods. Methods 9060A: Total Organic Carbon. SW-846. 3rd edition.

142

Vepraskas, M. J. and S. P. Faulkner. 2001. Redox chemistry of hydric soils. p. 85-105. In J. L. Richardson and M. J. Vepraskas (eds.) Wetland Soils: Genesis, Hydrology, Landscapes and Classification. Lewis Publishers, Boca Raton, FL, USA.

Wetzel, R. G. 2001. Limnology: Lake and River Ecosystems, Third Edition. Academic Press, Inc., San Francisco, CA, USA.

Wu, X. and W. J. Mitsch. 1998. Spatial and temporal patterns of algae in newly constructed freshwater wetlands. Wetlands 18: 9-20.

Zhang, L. and W.J. Mitsch. 2005. Modelling hydrological processes in created wetlands: An integrated system approach. Environmental Modelling & Software 20: 935-946.

143

Figure 5.1 Pumping system and water control structures for the two experimental wetlands at The Olentangy River Wetland Research Park (ORWRP) at The Ohio State University in Columbus, Ohio.

144

Figure 5.2 The 10-m grid and locations used for measuring sediment depth and collecting samples at the ORWRP experimental wetlands in May 2004. Shaded areas within the grid map represent approximate location of the deeper, open water (OW) zones.

145

a) 100 Wetland 1 Wetland 2

) 80 -2

Sediment (kg m (kg Sediment 20

0 EM OW Total

b) Wetland 1 ) 0.4 Wetland 2

0.3

0.2

0.1

0 Frequency (% of total area total of Frequency (% 0 102030405060708090100110 Sediment accumulation (kg m-2)

Figure 5.3 a) Mean (±1 SE) sediment accumulation at the emergent vegetation (EM) zones, open water (OW) zones and total wetland and; b) frequency distribution based on spatially interpolated data from kriging analyses for Wetland 1 and 2. Total wetland sediment accumulation is derived based on weighted average. Kriging data for Wetland 1 and 2 were based on anisotropic analyses (0◦ and 135◦, respectively; see text for variogram results and further details).

146

◦ 0 ◦ 45

◦ 90 135◦

Distance (10 m) Distance -2 Sediment (kg m ) > 80.0 > 70.0 > 60.0 > 50.0 > 40.0 > 30.0 > 20.0 > 10.0 0 0 8 8 16 Distance (10 m)

Figure 5.4 Spatial distribution maps of sediment accumulations for Wetland 1 and 2. Maps were generated by ordinary point kriging using anisotropic variogram models (0◦ and 135◦, respectively). Degree bearings provided for reference.

147

1 a)

Isotropic Anisotropic (0 degree) 0 Anisotropic (45 degree) Moran's I Anisotropic (135 degree)

-1 0 102030405060 Mean distance (m)

b) 1

Isotropic Anisotropic (0 degree) 0 Anisotropic (45 degree) Moran's I Anisotropic (135 degree)

-1 0 102030405060 Mean distance (m)

Figure 5.5 Correlogram (Moran’s I over mean distance) for sediment accumulation in a) Wetland 1 and b) Wetland 2 for isotropic and anisotropic (0◦, 45◦ and 135◦) analyses.

148

a) 3500 Wetland 1 Wetland 2 3000

) 2500 -2 2000

1500

Total C (g m C (g Total 1000

500

0 EM OW Total

b) 2500 Wetland 1 Wetland 2

) 2000 -2

1500

1000

Organic C m (g Organic 500

0 EM OW Total

1000 Wetland 1 c) Wetland 2 ) 800 -2

600

400

Inorganic C m (g 200

0 EM OW Total

Figure 5.6 Mean (±1 SE) accumulation of a) total C, b) organic C, and c) inorganic C for Wetland 1 and 2. Total wetland accumulation based on weighted average.

149

a) 300 Wetland 1 Wetland 2 250 ) -2 200 150 100 Total N (g m N (g Total 50 0 EM OW Total

b) 70 Wetland 1 Wetland 2 60 )

-2 50 40 30

Total P (g m 20 10 0 EM OW Total

c) 3000 Wetland 1 Wetland 2 2500 ) -2 2000

1500

1000 Total Ca (g m Ca (g Total 500

0 EM OW Total

Figure 5.7 Mean (±1 SE) accumulation of a) total N, b) total P, and c) total Ca for Wetland 1 and 2. Total wetland accumulation based on weighted average.

150

Wetland 1 Wetland 2 Sediment parameter OW Zone EM Zone Total OW Zone EM Zone Total Bulk density (g cm-3) 0.44 ±0.03 (11) 0.55 ±0.02 (22) 0.53 ±0.02 (33) 0.48 ±0.04 (16) 0.50 ±0.03 (20) 0.49 ±0.03 (36) Total C (%) 4.3 ±0.1 (5) 3.8 ±0.1 (17) 3.9 ±0.1 (22) 3.8 ±0.4 (6) 3.9 ±0.2 (16) 3.8 ±0.2 (24) Organic C (%) 2.9 ±0.1 (10) 3.6 ±0.1 (18) 3.5 ±0.1 (28) 2.9 ±0.1 (16) 3.9 ±0.2 (20) 3.7 ±0.2 (36) Inorganic C (%) 1.40 ±0.14 (4) 0.21 ±0.06 (9) 0.48 ±0.08 (10) 0.97 ±0.19 (4) 0.07 ±0.02 (8) 0.28 ±0.06 (12) Total N (%) 0.36 ±0.01 (5) 0.36 ±0.01 (17) 0.36 ±0.01 (22) 0.31 ±0.03 (6) 0.32 ±0.02 (19) 0.32 ±0.02 (25) Total P (ug g-1) 782 ±36 (3) 706 ±27 (7) 722 ±29 (10) 806 ±28 (4) 670 ±37 (7) 701 ±35 (11)

151 Total Ca (mg g-1) 47.2 ±6.1 (3) 5.0 ±0.5 (7) 14.3 ±1.7 (10) 34.4 ±6.4 (4) 5.4 ±0.5 (7) 12.1 ±1.9 (11) Values are means ±SE with sample size provided in parentheses. Total wetland values are weighted averages based on the relative proportion of OW and EM area at each wetland.

Table 5.1 Mean (±1 SE) physiochemical conditions of sediment in the emergent vegetation (EM) and open water (OW) zones of the planted (Wetland 1) and naturally colonized (Wetland 2) wetlands at the Olentangy River Wetland Research Park in May 2004.

151

Mean annual Sediment parameter accumulation rates Total sediment (kg m-2 yr-1) 4.5 - 4.9 Total C (g m-2 yr-1) 180.9 - 192.9 Organic C (g m-2 yr-1) 152.5 - 166.0 Inorganic C (g m-2 yr-1) 26.1 - 22.9 Total N (g m-2 yr-1) 16.2 - 16.6 Total P (g m-2 yr-1) 3.3 - 3.5 Total Ca (g m-2 yr-1) 80.8 - 86.3

Table 5.2 Range of mean annual accumulation rates of sediment and nutrients for Wetland 1 and 2 at the Olentangy River Wetland Research Park, 1994-2004.

152

REFERENCES

Ahn, C. and W.J. Mitsch. 2002. Scaling considerations of mesocosm wetlands in simulating large created freshwater marshes. Ecological Engineering 18: 327-342.

Alden, H. A. 1995. Hardwoods of North America. U. S. Department of Agriculture, Forest Service, Madison, WI, USA. FPL-GTR-83.

Anderson, C.J. and B.C. Cowell. 2004. Mulching effects on the seasonally flooded zone of west-central Florida, USA wetlands. Wetlands 24:811-819.

Anderson, C.J., W.J. Mitsch, and R.W. Nairn. 2005. Temporal and spatial development of surface soil conditions at two created riverine marshes. Journal of Environmental Quality 34(6): 2072-2081.

Anderson, K.L., K.A. Wheeler, J.A. Robinson, and O.H. Tuovinen. 2002. Atrazine mineralizaation potential in two wetlands. Water Research 36: 4785-4794.

AOAC Official Methods of Analysis. 1989. Method 990.03. Protein(crude) in Animal Feed Combustion Method (Dumas method). 17th edition 2002. Reference: JAOAC 72, 770 (1989).

Armentano, T. V. and G. M. Woodwell. 1975. Sedimentation rates in a Long Island marsh determined by 210Pb dating. Limnol. Oceanogr. 20:452-455.

153

Axt, J. R. and M. R. Walbridge. 1999. Phosphate removal capacity of palustrine forested wetlands and adjacent uplands in Virginia. Soil Sci. Soc. Am. J. 63:1019-1031.

Bayley, S. E., J. Zoltek, Jr., A. J. Hermann, T. J. Dolan and L. Tortora. 1985. Experimental manipulation of nutrients and water in a freshwater marsh: effects on biomass, decomposition, and nutrient accumulation. Limnology and Oceanography 30:500-512.

Bedford, B. L., M. R. Walbridge and A. Aldous. 1999. Patterns in nutrient availability and plant diversity of temperate North American wetlands. Ecology 80:2151-2169.

Bishel-Machung, L., R.P. Brooks, S.S. Yates, and K.L. Hoover. 1996. Soil properties of reference wetlands and wetland creation projects in Pennsylvania. Wetlands 16:532- 541.

Bolstad, P. V., J. M. Vose, and S. G. McNulty. 2001. Forest productivity, leaf area, and terrain in southern Appalachian deciduous forests. Forest Science 47:419-427.

Braskerud, B. C. 2001. The influence of vegetation on sedimentation and resuspension of soil particles in small constructed wetlands. J. Environ. Qual. 30:1447-1457.

Brown, S. L. 1981. A comparison of the structure, primary productivity, and transpiration of cypress ecosystems in Florida. Ecological Monographs 51:405-415.

Brown S. and D.L. Peterson. 1983. Structural characteristics and biomass production of two Illinois bottomland forests. American Midland Naturalist 110:107-117.

154

Brueske, C. C. and G. W. Barrett. 1994. Effects of vegetation and hydrologic load on sedimentation patterns in experimental wetland ecosystems. Ecological Engineering 3:429-447.

Bruland, G.L. and C.J. Richardson. 2004. A spatially explicit investigation of phosphorus sorption and related soil properties in two riparian wetlands. J. Environ. Qual. 33:785-794.

Camill, P. and J. S. Clark. 2000. Long-term perspectives on lagged ecosystem responses to climate change: permafrost in boreal peatlands and the grassland/woodland boundary. Ecosystems 3:534-544.

Campbell, D.A., C.A. Cole, and R.P. Brooks. 2002. A comparison of created and natural wetlands in Pennsylvania, USA. Wetlands Ecology and Management 10:41-49.

Casanova,M. T. and M. A. Brock. 2000. How do depth, duration and frequency of flooding influence the establishment of wetland plant communities? Plant Ecology 147:237-250.

Chadde, S. W. 2002. A Great Lakes Wetland Flora.- A complete guide to the aquatic and wetland plants of the upper Midwest. 2nd edition. PocketFlora Press, Laurium, MI, USA.

Cochran, M. 2001. Effect of Hydrology on bottomland hardwood forest productivity in central Ohio (USA). M.S. Thesis. The Ohio State University, Columbus, OH, USA.

Collins, M.E. and R.J. Kuehl. 2001. Organic matter accumulation and organic soils. In: J.L. Richardson and M.J. Vepraskas (eds.) Wetland Soils: Genesis, Hydrology, Landscapes and Classification. Lewis Publishers, Boca Raton, FL.

Conner, W. H. 1994. Effect of forest management practices on southern forested wetland productivity. Wetlands 14: 27-40.

Cooper, J. R. and J. W. Gilliam. 1987. Phosphorus redistribution from cultivated fields into riparian areas. Soil Sci. Soc. Am. J. 51:1600-1604.

Cooperrider, T. S., A. W. Cusick and J. T. Kartesz (eds.). 2001. Seventh Catalog of the Vascular Plants of Ohio. Ohio State University Press. Columbus, Ohio.

155

Craft, C.B. 1997. Dynamics of nitrogen and phosphorus retention during wetland ecosystem succession. Wetlands Ecology and Management 4:177-187.

Craft, C.B. 2001. Biology of wetland soils. In: J.L. Richardson and M.J. Vepraskas (eds.) Wetland Soils: Genesis, Hydrology, Landscapes and Classification. Lewis Publishers, Boca Raton, FL.

Craft, C. B. and W. P. Casey. 2000. Sediment and nutrient accumulation in floodplain and depressional freshwater wetlands of Georgia, USA. Wetlands 20: 323-332.

Craft, C. B., S. Broome, and C. Carlton. 2002. Fifteen years of vegetation and soil development after brackish-water marsh creation. Restoration Ecology 10: 248-258.

Crawford, R. M. M. 1992. Oxygen availability as an ecological limit to plant distribution. Advances in Ecological Research 23:93-185.

Cronk, J. K. and M. S. Fennessy. 2001. Wetland plants: biology and ecology. Lewis Publishers, Boca Raton, FL, USA. 462 p.

Darke, A. K. and J. P. Megonigal. 2003. Control of sediment deposition rates in two mid-Atlantic Coast tidal freshwater wetlands. Estuarine, Coastal and Shelf Science 57: 255-268.

Davis, S. M. 1991. Growth, decomposition, and nutrient retention of Cladium jamaicense Crantz and Typha domingensis Pers in Florida Everglades. Aquat. Bot. 40:203-224.

DeBusk, W.F., S. Newman, and K.R. Reddy. 2001. Spatio-temporal patterns of soil phosphorus enrichment in Everglades Water Conservation Area 2A. J. Environ. Qual. 30:1438-1446.

Dubbe, D. R, E. G. Graver, and D. C. Pratt. 1988. Production of cattail (Typha spp.) biomass in Minnesota, USA. Biomass 17:79-104.

156

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

Farnsworth, E. J. and L. A. Meyerson. 2003. Comparative ecophysiology of four wetland plant species along a continuum of invasiveness. Wetlands 23:750-762.

Fennessy, M. S., C. C. Brueske, and W. J. Mitsch. 1994. Sediment deposition patterns in restored freshwater wetlands using sediment traps. Ecological Engineering 3:409- 428.

Fennessy, M. S., J. K. Cronk and W. J. Mitsch. 1994. Macrophyte productivity and community development in created wetlands under experimental hydrological conditions. Ecological Engineering 3: 469-484.

Fennessy, M.S. and W.J. Mitsch. 2001. Effects of hydrology and spatial patterns of soil development in created riparian wetlands. Wetlands Ecology and Management 9:103-120.

Foster, J. R. 1992. Photosynthesis and water relations of the floodplain tree, boxelder (Acer negundo L.). Tree Physiol. 11:133-149.

Gamma Design Software. 2004. GS+ User’s Guide Version 7: GeoStatistics for the Environmental Sciences. Gamma Design Software, LLC. Plainwell, MI.

Garver, E. G., D. R. Dubbe, and D. C. Pratt. 1988. Seasonal patterns in accumulation and partitioning of biomass and macronutrients in Typha spp. Aquatic Botany 32:115- 127.

Gaudet, C. L. and P. A. Keddy. 1995. Competitive performance and species distribution in shoreline plant communities: a competitive approach. Ecology 76:280-291.

Gee, G.W. and J.W. Bauder. 1986. Particle-size analysis. p. 383–411. In A. Klute (ed.) Methods of soil analysis. Part 1. 2nd ed. Agron. Monogr. 9. ASA and SSSA, Madison, WI.

157

Giovannini, S. G. T. and D. M. L. Da Motta Marques. 1998. Establishment of three emergent macrophytes under different water regimes. Water Science & Technology 40:233-240.

Grace, J. B. and J. S. Harrison. 1986. The biology of Canadian Weeds. 73. Typha latifolia L. , Typha angustifolia L., and Typha xglauca Godr. Canadian Journal of Plant Science 66:361-379.

Grace, J.B. and R.G. Wetzel. 1981. Habitat partitioning and competitive displacement in cattails (Typha): experimental field studies. The American Naturalist 118:463-474.

Harter, S. K. and W. J. Mitsch. 2003. Patterns of short-term sedimentation in a freshwater created marsh. J. Environ. Qual. 32:325-334.

Hem, J.D. 1989. Study and Interpretation of the Chemical Characteristics of Natural Water, Third Edition. U.S. Geological Survey Water- Supply Paper 2254. Washington, DC.

Horppila, J. and L. Nurminen. 2001. The effect of an emergent macrophyte (Typha angustifolia) on sediment resuspension in a shallow north temperate lake. Freshwater Biology 46:1447-1455.

Hupp, C. R. and D. E. Bazemore. 1993. Temporal and spatial patterns of wetland sedimentation, West Tennessee. Journal of Hydrology 141:179-196.

Isaac, R.A. and W.A. Johnson. 1985. Elemental Analysis of Plant Tissue by Plasma Emission Spectroscopy: Collaborative Study. J. Assoc. Off. Anal. Chem. 68:499.

Isaaks, E.H. and R.M. Srivastava. 1989. An Introduction to Applied Geostatistics. Oxford University Press, Inc. New York, NY.

Johnson, F. L. and D. T. Bell. 1976. Tree growth and mortality in the streamside forest. Castanea 41:34-41.

158

Johnston, C. A. 1991. Sediment and nutrient retention by freshwater wetlands: effects on surface water quality. Critical Reviews in Environmental Control 21:491-565.

Johnston, C. A., S. D. Bridgham, and J. P. Schubauer-Berigan. 2001. Nutrient dynamics in relation to geomorphology of riverine wetlands. Soil Sci. Soc. Am. J. 65:557-577.

Jones, T. A. and S. C. Thomas. 2004. The time course of diameter increment to selection harvests in Acer saccharum. Can. J. Res./Rev. Can. Rech. For. 34:1525- 1533.

Junk, W. J., P. B. Bayley, and R. E. Sparks. 1989. The flood pulse concept in river- floodplain systems. In: D. P. Dodge (ed.). Proceedings of the International Large River Symposium. Special Issue of Journal of Canadian Fisheries and Aquatic Sciences 106:11-127.

Jurinak, J. J. and R. A. Griffin. 1992. Nitrate ion adsorption by calcium carbonate. Soil Sci. 113: 130-135.

Kang, H., C. Freeman, D. Lee, and W.J. Mitsch. 1998. Enzyme activities in constructed wetlands: Implication for water quality amelioration. Hydrobiologia 368:231-235.

Kimmins, J. P. 1987. Forest ecology. Macmillan Publishing Company. New York, NY, USA.

Koerselman, W. and A. F. M. Meuleman. 1996. The vegetation N:P ratio: a new tool to detect the nature of nutrient limitation. The Journal of Applied Ecology 33:1441- 1450.

Koreny, J.S., W.J. Mitsch, E.S. Bair, and X. Wu. 1999. Regional and local hydrology of a created riparian wetland system. Wetlands 19:182-193.

Kozlowski, T. T. 1997. Responses of woody plants to flooding and salinity. Tree Physiology Monograph 1:1-17.

159

Kozlowski, T. T. and S. G. Pallardy. 2002. Acclimation and adaptive responses of woody plants to environmental stresses. The Botanical Review 68:270-334.

Kuo, S. 1996. Phosphorus. p. 894-895. In: Methods of Soil Analysis, Part 3—Chemical Methods. Soil Science Society of America, Madison, WI.

Kvet, J. and S. Husak. 1978. Primary data on biomass and production estimates in typical stands offishpond littoral plant communities. p.211-216. In: Pond Littoral Ecosystems, D. Dykyjova and J. Kvet (eds.) Springer-Verlag, Berlin.